12345...102030...


Artificial intelligence – Wikipedia

Intelligence demonstrated by machines

Artificial intelligence (AI), sometimes called machine intelligence, is intelligence demonstrated by machines, in contrast to the natural intelligence displayed by humans and other animals. In computer science AI research is defined as the study of “intelligent agents”: any device that perceives its environment and takes actions that maximize its chance of successfully achieving its goals.[1] Colloquially, the term “artificial intelligence” is applied when a machine mimics “cognitive” functions that humans associate with other human minds, such as “learning” and “problem solving”.

The scope of AI is disputed: as machines become increasingly capable, tasks considered as requiring “intelligence” are often removed from the definition, a phenomenon known as the AI effect, leading to the quip in Tesler’s Theorem, “AI is whatever hasn’t been done yet.”[3] For instance, optical character recognition is frequently excluded from “artificial intelligence”, having become a routine technology.[4] Modern machine capabilities generally classified as AI include successfully understanding human speech, competing at the highest level in strategic game systems (such as chess and Go),[6] autonomously operating cars, and intelligent routing in content delivery networks and military simulations.

Artificial intelligence was founded as an academic discipline in 1956, and in the years since has experienced several waves of optimism,[7][8] followed by disappointment and the loss of funding (known as an “AI winter”),[9][10] followed by new approaches, success and renewed funding.[8][11] For most of its history, AI research has been divided into subfields that often fail to communicate with each other.[12] These sub-fields are based on technical considerations, such as particular goals (e.g. “robotics” or “machine learning”),[13] the use of particular tools (“logic” or artificial neural networks), or deep philosophical differences.[14][15][16] Subfields have also been based on social factors (particular institutions or the work of particular researchers).[12]

The traditional problems (or goals) of AI research include reasoning, knowledge representation, planning, learning, natural language processing, perception and the ability to move and manipulate objects.[13] General intelligence is among the field’s long-term goals.[17] Approaches include statistical methods, computational intelligence, and traditional symbolic AI. Many tools are used in AI, including versions of search and mathematical optimization, artificial neural networks, and methods based on statistics, probability and economics. The AI field draws upon computer science, information engineering, mathematics, psychology, linguistics, philosophy, and many others.

The field was founded on the claim that human intelligence “can be so precisely described that a machine can be made to simulate it”.[18] This raises philosophical arguments about the nature of the mind and the ethics of creating artificial beings endowed with human-like intelligence which are issues that have been explored by myth, fiction and philosophy since antiquity.[19] Some people also consider AI to be a danger to humanity if it progresses unabated.[20] Others believe that AI, unlike previous technological revolutions, will create a risk of mass unemployment.[21]However Google is hosting a global contest to develop AI thats beneficial for humanity[22]

In the twenty-first century, AI techniques have experienced a resurgence following concurrent advances in computer power, large amounts of data, and theoretical understanding; and AI techniques have become an essential part of the technology industry, helping to solve many challenging problems in computer science, software engineering and operations research.[23][11]

Thought-capable artificial beings appeared as storytelling devices in antiquity,[24] and have been common in fiction, as in Mary Shelley’s Frankenstein or Karel apek’s R.U.R. (Rossum’s Universal Robots).[25] These characters and their fates raised many of the same issues now discussed in the ethics of artificial intelligence.[19]

The study of mechanical or “formal” reasoning began with philosophers and mathematicians in antiquity. The study of mathematical logic led directly to Alan Turing’s theory of computation, which suggested that a machine, by shuffling symbols as simple as “0” and “1”, could simulate any conceivable act of mathematical deduction. This insight, that digital computers can simulate any process of formal reasoning, is known as the ChurchTuring thesis.[26] Along with concurrent discoveries in neurobiology, information theory and cybernetics, this led researchers to consider the possibility of building an electronic brain. Turing proposed that “if a human could not distinguish between responses from a machine and a human, the machine could be considered intelligent”.[27] The first work that is now generally recognized as AI was McCullouch and Pitts’ 1943 formal design for Turing-complete “artificial neurons”.

The field of AI research was born at a workshop at Dartmouth College in 1956.[29] Attendees Allen Newell (CMU), Herbert Simon (CMU), John McCarthy (MIT), Marvin Minsky (MIT) and Arthur Samuel (IBM) became the founders and leaders of AI research.[30] They and their students produced programs that the press described as “astonishing”: computers were learning checkers strategies (c. 1954)[32] (and by 1959 were reportedly playing better than the average human),[33] solving word problems in algebra, proving logical theorems (Logic Theorist, first run c. 1956) and speaking English.[34] By the middle of the 1960s, research in the U.S. was heavily funded by the Department of Defense[35] and laboratories had been established around the world.[36] AI’s founders were optimistic about the future: Herbert Simon predicted, “machines will be capable, within twenty years, of doing any work a man can do”. Marvin Minsky agreed, writing, “within a generation… the problem of creating ‘artificial intelligence’ will substantially be solved”.[7]

They failed to recognize the difficulty of some of the remaining tasks. Progress slowed and in 1974, in response to the criticism of Sir James Lighthill and ongoing pressure from the US Congress to fund more productive projects, both the U.S. and British governments cut off exploratory research in AI. The next few years would later be called an “AI winter”,[9] a period when obtaining funding for AI projects was difficult.

In the early 1980s, AI research was revived by the commercial success of expert systems,[38] a form of AI program that simulated the knowledge and analytical skills of human experts. By 1985, the market for AI had reached over a billion dollars. At the same time, Japan’s fifth generation computer project inspired the U.S and British governments to restore funding for academic research.[8] However, beginning with the collapse of the Lisp Machine market in 1987, AI once again fell into disrepute, and a second, longer-lasting hiatus began.[10]

In the late 1990s and early 21st century, AI began to be used for logistics, data mining, medical diagnosis and other areas.[23] The success was due to increasing computational power (see Moore’s law), greater emphasis on solving specific problems, new ties between AI and other fields (such as statistics, economics and mathematics), and a commitment by researchers to mathematical methods and scientific standards.[39] Deep Blue became the first computer chess-playing system to beat a reigning world chess champion, Garry Kasparov, on 11 May 1997.

In 2011, a Jeopardy! quiz show exhibition match, IBM’s question answering system, Watson, defeated the two greatest Jeopardy! champions, Brad Rutter and Ken Jennings, by a significant margin. Faster computers, algorithmic improvements, and access to large amounts of data enabled advances in machine learning and perception; data-hungry deep learning methods started to dominate accuracy benchmarks around 2012.[42] The Kinect, which provides a 3D bodymotion interface for the Xbox 360 and the Xbox One, uses algorithms that emerged from lengthy AI research[43] as do intelligent personal assistants in smartphones.[44] In March 2016, AlphaGo won 4 out of 5 games of Go in a match with Go champion Lee Sedol, becoming the first computer Go-playing system to beat a professional Go player without handicaps.[6][45] In the 2017 Future of Go Summit, AlphaGo won a three-game match with Ke Jie,[46] who at the time continuously held the world No. 1 ranking for two years.[47][48] This marked the completion of a significant milestone in the development of Artificial Intelligence as Go is an extremely complex game, more so than Chess.

According to Bloomberg’s Jack Clark, 2015 was a landmark year for artificial intelligence, with the number of software projects that use AI within Google increased from a “sporadic usage” in 2012 to more than 2,700 projects. Clark also presents factual data indicating that error rates in image processing tasks have fallen significantly since 2011.[49] He attributes this to an increase in affordable neural networks, due to a rise in cloud computing infrastructure and to an increase in research tools and datasets.[11] Other cited examples include Microsoft’s development of a Skype system that can automatically translate from one language to another and Facebook’s system that can describe images to blind people.[49] In a 2017 survey, one in five companies reported they had “incorporated AI in some offerings or processes”.[50][51] Around 2016, China greatly accelerated its government funding; given its large supply of data and its rapidly increasing research output, some observers believe it may be on track to becoming an “AI superpower”.[52][53]

A typical AI perceives its environment and takes actions that maximize its chance of successfully achieving its goals.[1] An AI’s intended goal function can be simple (“1 if the AI wins a game of Go, 0 otherwise”) or complex (“Do actions mathematically similar to the actions that got you rewards in the past”). Goals can be explicitly defined, or can be induced. If the AI is programmed for “reinforcement learning”, goals can be implicitly induced by rewarding some types of behavior and punishing others.[a] Alternatively, an evolutionary system can induce goals by using a “fitness function” to mutate and preferentially replicate high-scoring AI systems; this is similar to how animals evolved to innately desire certain goals such as finding food, or how dogs can be bred via artificial selection to possess desired traits. Some AI systems, such as nearest-neighbor, instead reason by analogy; these systems are not generally given goals, except to the degree that goals are somehow implicit in their training data. Such systems can still be benchmarked if the non-goal system is framed as a system whose “goal” is to successfully accomplish its narrow classification task.[56]

AI often revolves around the use of algorithms. An algorithm is a set of unambiguous instructions that a mechanical computer can execute.[b] A complex algorithm is often built on top of other, simpler, algorithms. A simple example of an algorithm is the following recipe for optimal play at tic-tac-toe:

Many AI algorithms are capable of learning from data; they can enhance themselves by learning new heuristics (strategies, or “rules of thumb”, that have worked well in the past), or can themselves write other algorithms. Some of the “learners” described below, including Bayesian networks, decision trees, and nearest-neighbor, could theoretically, if given infinite data, time, and memory, learn to approximate any function, including whatever combination of mathematical functions would best describe the entire world. These learners could therefore, in theory, derive all possible knowledge, by considering every possible hypothesis and matching it against the data. In practice, it is almost never possible to consider every possibility, because of the phenomenon of “combinatorial explosion”, where the amount of time needed to solve a problem grows exponentially. Much of AI research involves figuring out how to identify and avoid considering broad swaths of possibilities that are unlikely to be fruitful.[58] For example, when viewing a map and looking for the shortest driving route from Denver to New York in the East, one can in most cases skip looking at any path through San Francisco or other areas far to the West; thus, an AI wielding an pathfinding algorithm like A* can avoid the combinatorial explosion that would ensue if every possible route had to be ponderously considered in turn.[60]

The earliest (and easiest to understand) approach to AI was symbolism (such as formal logic): “If an otherwise healthy adult has a fever, then they may have influenza”. A second, more general, approach is Bayesian inference: “If the current patient has a fever, adjust the probability they have influenza in such-and-such way”. The third major approach, extremely popular in routine business AI applications, are analogizers such as SVM and nearest-neighbor: “After examining the records of known past patients whose temperature, symptoms, age, and other factors mostly match the current patient, X% of those patients turned out to have influenza”. A fourth approach is harder to intuitively understand, but is inspired by how the brain’s machinery works: the artificial neural network approach uses artificial “neurons” that can learn by comparing itself to the desired output and altering the strengths of the connections between its internal neurons to “reinforce” connections that seemed to be useful. These four main approaches can overlap with each other and with evolutionary systems; for example, neural nets can learn to make inferences, to generalize, and to make analogies. Some systems implicitly or explicitly use multiple of these approaches, alongside many other AI and non-AI algorithms;[61] the best approach is often different depending on the problem.[63]

Learning algorithms work on the basis that strategies, algorithms, and inferences that worked well in the past are likely to continue working well in the future. These inferences can be obvious, such as “since the sun rose every morning for the last 10,000 days, it will probably rise tomorrow morning as well”. They can be nuanced, such as “X% of families have geographically separate species with color variants, so there is an Y% chance that undiscovered black swans exist”. Learners also work on the basis of “Occam’s razor”: The simplest theory that explains the data is the likeliest. Therefore, to be successful, a learner must be designed such that it prefers simpler theories to complex theories, except in cases where the complex theory is proven substantially better. Settling on a bad, overly complex theory gerrymandered to fit all the past training data is known as overfitting. Many systems attempt to reduce overfitting by rewarding a theory in accordance with how well it fits the data, but penalizing the theory in accordance with how complex the theory is. Besides classic overfitting, learners can also disappoint by “learning the wrong lesson”. A toy example is that an image classifier trained only on pictures of brown horses and black cats might conclude that all brown patches are likely to be horses. A real-world example is that, unlike humans, current image classifiers don’t determine the spatial relationship between components of the picture; instead, they learn abstract patterns of pixels that humans are oblivious to, but that linearly correlate with images of certain types of real objects. Faintly superimposing such a pattern on a legitimate image results in an “adversarial” image that the system misclassifies.[c][66][67][68]

Compared with humans, existing AI lacks several features of human “commonsense reasoning”; most notably, humans have powerful mechanisms for reasoning about “nave physics” such as space, time, and physical interactions. This enables even young children to easily make inferences like “If I roll this pen off a table, it will fall on the floor”. Humans also have a powerful mechanism of “folk psychology” that helps them to interpret natural-language sentences such as “The city councilmen refused the demonstrators a permit because they advocated violence”. (A generic AI has difficulty inferring whether the councilmen or the demonstrators are the ones alleged to be advocating violence.)[71][72][73] This lack of “common knowledge” means that AI often makes different mistakes than humans make, in ways that can seem incomprehensible. For example, existing self-driving cars cannot reason about the location nor the intentions of pedestrians in the exact way that humans do, and instead must use non-human modes of reasoning to avoid accidents.[74][75][76]

The overall research goal of artificial intelligence is to create technology that allows computers and machines to function in an intelligent manner. The general problem of simulating (or creating) intelligence has been broken down into sub-problems. These consist of particular traits or capabilities that researchers expect an intelligent system to display. The traits described below have received the most attention.[13]

Early researchers developed algorithms that imitated step-by-step reasoning that humans use when they solve puzzles or make logical deductions.[77] By the late 1980s and 1990s, AI research had developed methods for dealing with uncertain or incomplete information, employing concepts from probability and economics.[78]

These algorithms proved to be insufficient for solving large reasoning problems, because they experienced a “combinatorial explosion”: they became exponentially slower as the problems grew larger.[58] In fact, even humans rarely use the step-by-step deduction that early AI research was able to model. They solve most of their problems using fast, intuitive judgements.[79]

Knowledge representation[80] and knowledge engineering[81] are central to classical AI research. Some “expert systems” attempt to gather together explicit knowledge possessed by experts in some narrow domain. In addition, some projects attempt to gather the “commonsense knowledge” known to the average person into a database containing extensive knowledge about the world. Among the things a comprehensive commonsense knowledge base would contain are: objects, properties, categories and relations between objects;[82] situations, events, states and time;[83] causes and effects;[84] knowledge about knowledge (what we know about what other people know);[85] and many other, less well researched domains. A representation of “what exists” is an ontology: the set of objects, relations, concepts, and properties formally described so that software agents can interpret them. The semantics of these are captured as description logic concepts, roles, and individuals, and typically implemented as classes, properties, and individuals in the Web Ontology Language.[86] The most general ontologies are called upper ontologies, which attempt to provide a foundation for all other knowledge[87] by acting as mediators between domain ontologies that cover specific knowledge about a particular knowledge domain (field of interest or area of concern). Such formal knowledge representations can be used in content-based indexing and retrieval,[88] scene interpretation,[89] clinical decision support,[90] knowledge discovery (mining “interesting” and actionable inferences from large databases),[91] and other areas.[92]

Among the most difficult problems in knowledge representation are:

Intelligent agents must be able to set goals and achieve them.[99] They need a way to visualize the futurea representation of the state of the world and be able to make predictions about how their actions will change itand be able to make choices that maximize the utility (or “value”) of available choices.[100]

In classical planning problems, the agent can assume that it is the only system acting in the world, allowing the agent to be certain of the consequences of its actions.[101] However, if the agent is not the only actor, then it requires that the agent can reason under uncertainty. This calls for an agent that can not only assess its environment and make predictions, but also evaluate its predictions and adapt based on its assessment.[102]

Multi-agent planning uses the cooperation and competition of many agents to achieve a given goal. Emergent behavior such as this is used by evolutionary algorithms and swarm intelligence.[103]

Machine learning, a fundamental concept of AI research since the field’s inception,[104] is the study of computer algorithms that improve automatically through experience.[105][106]

Unsupervised learning is the ability to find patterns in a stream of input, without requiring a human to label the inputs first.[107] Supervised learning includes both classification and numerical regression, which requires a human to label the input data first. Classification is used to determine what category something belongs in, after seeing a number of examples of things from several categories. Regression is the attempt to produce a function that describes the relationship between inputs and outputs and predicts how the outputs should change as the inputs change.[106] Both classifiers and regression learners can be viewed as “function approximators” trying to learn an unknown (possibly implicit) function; for example, a spam classifier can be viewed as learning a function that maps from the text of an email to one of two categories, “spam” or “not spam”. Computational learning theory can assess learners by computational complexity, by sample complexity (how much data is required), or by other notions of optimization.[108] In reinforcement learning[109] the agent is rewarded for good responses and punished for bad ones. The agent uses this sequence of rewards and punishments to form a strategy for operating in its problem space.

Natural language processing[110] (NLP) gives machines the ability to read and understand human language. A sufficiently powerful natural language processing system would enable natural-language user interfaces and the acquisition of knowledge directly from human-written sources, such as newswire texts. Some straightforward applications of natural language processing include information retrieval, text mining, question answering[111] and machine translation.[112] Many current approaches use word co-occurrence frequencies to construct syntactic representations of text. “Keyword spotting” strategies for search are popular and scalable but dumb; a search query for “dog” might only match documents with the literal word “dog” and miss a document with the word “poodle”. “Lexical affinity” strategies use the occurrence of words such as “accident” to assess the sentiment of a document. Modern statistical NLP approaches can combine all these strategies as well as others, and often achieve acceptable accuracy at the page or paragraph level, but continue to lack the semantic understanding required to classify isolated sentences well. Besides the usual difficulties with encoding semantic commonsense knowledge, existing semantic NLP sometimes scales too poorly to be viable in business applications. Beyond semantic NLP, the ultimate goal of “narrative” NLP is to embody a full understanding of commonsense reasoning.[113]

Machine perception[114] is the ability to use input from sensors (such as cameras (visible spectrum or infrared), microphones, wireless signals, and active lidar, sonar, radar, and tactile sensors) to deduce aspects of the world. Applications include speech recognition,[115] facial recognition, and object recognition.[116] Computer vision is the ability to analyze visual input. Such input is usually ambiguous; a giant, fifty-meter-tall pedestrian far away may produce exactly the same pixels as a nearby normal-sized pedestrian, requiring the AI to judge the relative likelihood and reasonableness of different interpretations, for example by using its “object model” to assess that fifty-meter pedestrians do not exist.[117]

AI is heavily used in robotics.[118] Advanced robotic arms and other industrial robots, widely used in modern factories, can learn from experience how to move efficiently despite the presence of friction and gear slippage.[119] A modern mobile robot, when given a small, static, and visible environment, can easily determine its location and map its environment; however, dynamic environments, such as (in endoscopy) the interior of a patient’s breathing body, pose a greater challenge. Motion planning is the process of breaking down a movement task into “primitives” such as individual joint movements. Such movement often involves compliant motion, a process where movement requires maintaining physical contact with an object.[121][122] Moravec’s paradox generalizes that low-level sensorimotor skills that humans take for granted are, counterintuitively, difficult to program into a robot; the paradox is named after Hans Moravec, who stated in 1988 that “it is comparatively easy to make computers exhibit adult level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility”.[123][124] This is attributed to the fact that, unlike checkers, physical dexterity has been a direct target of natural selection for millions of years.[125]

Moravec’s paradox can be extended to many forms of social intelligence.[127][128] Distributed multi-agent coordination of autonomous vehicles remains a difficult problem.[129] Affective computing is an interdisciplinary umbrella that comprises systems which recognize, interpret, process, or simulate human affects. Moderate successes related to affective computing include textual sentiment analysis and, more recently, multimodal affect analysis (see multimodal sentiment analysis), wherein AI classifies the affects displayed by a videotaped subject.[133]

In the long run, social skills and an understanding of human emotion and game theory would be valuable to a social agent. Being able to predict the actions of others by understanding their motives and emotional states would allow an agent to make better decisions. Some computer systems mimic human emotion and expressions to appear more sensitive to the emotional dynamics of human interaction, or to otherwise facilitate humancomputer interaction.[134] Similarly, some virtual assistants are programmed to speak conversationally or even to banter humorously; this tends to give nave users an unrealistic conception of how intelligent existing computer agents actually are.[135]

Historically, projects such as the Cyc knowledge base (1984) and the massive Japanese Fifth Generation Computer Systems initiative (19821992) attempted to cover the breadth of human cognition. These early projects failed to escape the limitations of non-quantitative symbolic logic models and, in retrospect, greatly underestimated the difficulty of cross-domain AI. Nowadays, the vast majority of current AI researchers work instead on tractable “narrow AI” applications (such as medical diagnosis or automobile navigation).[136] Many researchers predict that such “narrow AI” work in different individual domains will eventually be incorporated into a machine with artificial general intelligence (AGI), combining most of the narrow skills mentioned in this article and at some point even exceeding human ability in most or all these areas.[17][137] Many advances have general, cross-domain significance. One high-profile example is that DeepMind in the 2010s developed a “generalized artificial intelligence” that could learn many diverse Atari games on its own, and later developed a variant of the system which succeeds at sequential learning.[138][139][140] Besides transfer learning,[141] hypothetical AGI breakthroughs could include the development of reflective architectures that can engage in decision-theoretic metareasoning, and figuring out how to “slurp up” a comprehensive knowledge base from the entire unstructured Web. Some argue that some kind of (currently-undiscovered) conceptually straightforward, but mathematically difficult, “Master Algorithm” could lead to AGI. Finally, a few “emergent” approaches look to simulating human intelligence extremely closely, and believe that anthropomorphic features like an artificial brain or simulated child development may someday reach a critical point where general intelligence emerges.[143][144]

Many of the problems in this article may also require general intelligence, if machines are to solve the problems as well as people do. For example, even specific straightforward tasks, like machine translation, require that a machine read and write in both languages (NLP), follow the author’s argument (reason), know what is being talked about (knowledge), and faithfully reproduce the author’s original intent (social intelligence). A problem like machine translation is considered “AI-complete”, because all of these problems need to be solved simultaneously in order to reach human-level machine performance.

There is no established unifying theory or paradigm that guides AI research. Researchers disagree about many issues.[145] A few of the most long standing questions that have remained unanswered are these: should artificial intelligence simulate natural intelligence by studying psychology or neurobiology? Or is human biology as irrelevant to AI research as bird biology is to aeronautical engineering?[14]Can intelligent behavior be described using simple, elegant principles (such as logic or optimization)? Or does it necessarily require solving a large number of completely unrelated problems?[15]

In the 1940s and 1950s, a number of researchers explored the connection between neurobiology, information theory, and cybernetics. Some of them built machines that used electronic networks to exhibit rudimentary intelligence, such as W. Grey Walter’s turtles and the Johns Hopkins Beast. Many of these researchers gathered for meetings of the Teleological Society at Princeton University and the Ratio Club in England.[146] By 1960, this approach was largely abandoned, although elements of it would be revived in the 1980s.

When access to digital computers became possible in the middle 1950s, AI research began to explore the possibility that human intelligence could be reduced to symbol manipulation. The research was centered in three institutions: Carnegie Mellon University, Stanford and MIT, and as described below, each one developed its own style of research. John Haugeland named these symbolic approaches to AI “good old fashioned AI” or “GOFAI”.[147] During the 1960s, symbolic approaches had achieved great success at simulating high-level thinking in small demonstration programs. Approaches based on cybernetics or artificial neural networks were abandoned or pushed into the background.[148]Researchers in the 1960s and the 1970s were convinced that symbolic approaches would eventually succeed in creating a machine with artificial general intelligence and considered this the goal of their field.

Economist Herbert Simon and Allen Newell studied human problem-solving skills and attempted to formalize them, and their work laid the foundations of the field of artificial intelligence, as well as cognitive science, operations research and management science. Their research team used the results of psychological experiments to develop programs that simulated the techniques that people used to solve problems. This tradition, centered at Carnegie Mellon University would eventually culminate in the development of the Soar architecture in the middle 1980s.[149][150]

Unlike Simon and Newell, John McCarthy felt that machines did not need to simulate human thought, but should instead try to find the essence of abstract reasoning and problem-solving, regardless of whether people used the same algorithms.[14] His laboratory at Stanford (SAIL) focused on using formal logic to solve a wide variety of problems, including knowledge representation, planning and learning.[151] Logic was also the focus of the work at the University of Edinburgh and elsewhere in Europe which led to the development of the programming language Prolog and the science of logic programming.[152]

Researchers at MIT (such as Marvin Minsky and Seymour Papert)[153] found that solving difficult problems in vision and natural language processing required ad-hoc solutions they argued that there was no simple and general principle (like logic) that would capture all the aspects of intelligent behavior. Roger Schank described their “anti-logic” approaches as “scruffy” (as opposed to the “neat” paradigms at CMU and Stanford).[15] Commonsense knowledge bases (such as Doug Lenat’s Cyc) are an example of “scruffy” AI, since they must be built by hand, one complicated concept at a time.[154]

When computers with large memories became available around 1970, researchers from all three traditions began to build knowledge into AI applications.[155] This “knowledge revolution” led to the development and deployment of expert systems (introduced by Edward Feigenbaum), the first truly successful form of AI software.[38] Key component on system arhitecute for all expert systems is Knowledge base, which stores facts and rules that illustrates AI.[156] The knowledge revolution was also driven by the realization that enormous amounts of knowledge would be required by many simple AI applications.

By the 1980s, progress in symbolic AI seemed to stall and many believed that symbolic systems would never be able to imitate all the processes of human cognition, especially perception, robotics, learning and pattern recognition. A number of researchers began to look into “sub-symbolic” approaches to specific AI problems.[16] Sub-symbolic methods manage to approach intelligence without specific representations of knowledge.

This includes embodied, situated, behavior-based, and nouvelle AI. Researchers from the related field of robotics, such as Rodney Brooks, rejected symbolic AI and focused on the basic engineering problems that would allow robots to move and survive.[157] Their work revived the non-symbolic viewpoint of the early cybernetics researchers of the 1950s and reintroduced the use of control theory in AI. This coincided with the development of the embodied mind thesis in the related field of cognitive science: the idea that aspects of the body (such as movement, perception and visualization) are required for higher intelligence.

Within developmental robotics, developmental learning approaches are elaborated upon to allow robots to accumulate repertoires of novel skills through autonomous self-exploration, social interaction with human teachers, and the use of guidance mechanisms (active learning, maturation, motor synergies, etc.).[158][159]

Interest in neural networks and “connectionism” was revived by David Rumelhart and others in the middle of the 1980s.[162] Artificial neural networks are an example of soft computing — they are solutions to problems which cannot be solved with complete logical certainty, and where an approximate solution is often sufficient. Other soft computing approaches to AI include fuzzy systems, evolutionary computation and many statistical tools. The application of soft computing to AI is studied collectively by the emerging discipline of computational intelligence.[163]

Much of traditional GOFAI got bogged down on ad hoc patches to symbolic computation that worked on their own toy models but failed to generalize to real-world results. However, around the 1990s, AI researchers adopted sophisticated mathematical tools, such as hidden Markov models (HMM), information theory, and normative Bayesian decision theory to compare or to unify competing architectures. The shared mathematical language permitted a high level of collaboration with more established fields (like mathematics, economics or operations research).[d] Compared with GOFAI, new “statistical learning” techniques such as HMM and neural networks were gaining higher levels of accuracy in many practical domains such as data mining, without necessarily acquiring semantic understanding of the datasets. The increased successes with real-world data led to increasing emphasis on comparing different approaches against shared test data to see which approach performed best in a broader context than that provided by idiosyncratic toy models; AI research was becoming more scientific. Nowadays results of experiments are often rigorously measurable, and are sometimes (with difficulty) reproducible.[39][164] Different statistical learning techniques have different limitations; for example, basic HMM cannot model the infinite possible combinations of natural language. Critics note that the shift from GOFAI to statistical learning is often also a shift away from Explainable AI. In AGI research, some scholars caution against over-reliance on statistical learning, and argue that continuing research into GOFAI will still be necessary to attain general intelligence.

AI has developed a large number of tools to solve the most difficult problems in computer science. A few of the most general of these methods are discussed below.

Many problems in AI can be solved in theory by intelligently searching through many possible solutions:[173] Reasoning can be reduced to performing a search. For example, logical proof can be viewed as searching for a path that leads from premises to conclusions, where each step is the application of an inference rule.[174] Planning algorithms search through trees of goals and subgoals, attempting to find a path to a target goal, a process called means-ends analysis.[175] Robotics algorithms for moving limbs and grasping objects use local searches in configuration space.[119] Many learning algorithms use search algorithms based on optimization.

Simple exhaustive searches[176] are rarely sufficient for most real-world problems: the search space (the number of places to search) quickly grows to astronomical numbers. The result is a search that is too slow or never completes. The solution, for many problems, is to use “heuristics” or “rules of thumb” that prioritize choices in favor of those that are more likely to reach a goal and to do so in a shorter number of steps. In some search methodologies heuristics can also serve to entirely eliminate some choices that are unlikely to lead to a goal (called “pruning the search tree”). Heuristics supply the program with a “best guess” for the path on which the solution lies.[177] Heuristics limit the search for solutions into a smaller sample size.

A very different kind of search came to prominence in the 1990s, based on the mathematical theory of optimization. For many problems, it is possible to begin the search with some form of a guess and then refine the guess incrementally until no more refinements can be made. These algorithms can be visualized as blind hill climbing: we begin the search at a random point on the landscape, and then, by jumps or steps, we keep moving our guess uphill, until we reach the top. Other optimization algorithms are simulated annealing, beam search and random optimization.[178]

Evolutionary computation uses a form of optimization search. For example, they may begin with a population of organisms (the guesses) and then allow them to mutate and recombine, selecting only the fittest to survive each generation (refining the guesses). Classic evolutionary algorithms include genetic algorithms, gene expression programming, and genetic programming.[179] Alternatively, distributed search processes can coordinate via swarm intelligence algorithms. Two popular swarm algorithms used in search are particle swarm optimization (inspired by bird flocking) and ant colony optimization (inspired by ant trails).[180][181]

Logic[182] is used for knowledge representation and problem solving, but it can be applied to other problems as well. For example, the satplan algorithm uses logic for planning[183] and inductive logic programming is a method for learning.[184]

Several different forms of logic are used in AI research. Propositional logic[185] involves truth functions such as “or” and “not”. First-order logic[186] adds quantifiers and predicates, and can express facts about objects, their properties, and their relations with each other. Fuzzy set theory assigns a “degree of truth” (between 0 and 1) to vague statements such as “Alice is old” (or rich, or tall, or hungry) that are too linguistically imprecise to be completely true or false. Fuzzy logic is successfully used in control systems to allow experts to contribute vague rules such as “if you are close to the destination station and moving fast, increase the train’s brake pressure”; these vague rules can then be numerically refined within the system. Fuzzy logic fails to scale well in knowledge bases; many AI researchers question the validity of chaining fuzzy-logic inferences.[e][188][189]

Default logics, non-monotonic logics and circumscription[94] are forms of logic designed to help with default reasoning and the qualification problem. Several extensions of logic have been designed to handle specific domains of knowledge, such as: description logics;[82] situation calculus, event calculus and fluent calculus (for representing events and time);[83] causal calculus;[84] belief calculus;[190] and modal logics.[85]

Overall, qualitiative symbolic logic is brittle and scales poorly in the presence of noise or other uncertainty. Exceptions to rules are numerous, and it is difficult for logical systems to function in the presence of contradictory rules.[192]

Many problems in AI (in reasoning, planning, learning, perception, and robotics) require the agent to operate with incomplete or uncertain information. AI researchers have devised a number of powerful tools to solve these problems using methods from probability theory and economics.[193]

Bayesian networks[194] are a very general tool that can be used for a large number of problems: reasoning (using the Bayesian inference algorithm),[195] learning (using the expectation-maximization algorithm),[f][197] planning (using decision networks)[198] and perception (using dynamic Bayesian networks).[199] Probabilistic algorithms can also be used for filtering, prediction, smoothing and finding explanations for streams of data, helping perception systems to analyze processes that occur over time (e.g., hidden Markov models or Kalman filters).[199] Compared with symbolic logic, formal Bayesian inference is computationally expensive. For inference to be tractable, most observations must be conditionally independent of one another. Complicated graphs with diamonds or other “loops” (undirected cycles) can require a sophisticated method such as Markov Chain Monte Carlo, which spreads an ensemble of random walkers throughout the Bayesian network and attempts to converge to an assessment of the conditional probabilities. Bayesian networks are used on Xbox Live to rate and match players; wins and losses are “evidence” of how good a player is. AdSense uses a Bayesian network with over 300 million edges to learn which ads to serve.

A key concept from the science of economics is “utility”: a measure of how valuable something is to an intelligent agent. Precise mathematical tools have been developed that analyze how an agent can make choices and plan, using decision theory, decision analysis,[200] and information value theory.[100] These tools include models such as Markov decision processes,[201] dynamic decision networks,[199] game theory and mechanism design.[202]

The simplest AI applications can be divided into two types: classifiers (“if shiny then diamond”) and controllers (“if shiny then pick up”). Controllers do, however, also classify conditions before inferring actions, and therefore classification forms a central part of many AI systems. Classifiers are functions that use pattern matching to determine a closest match. They can be tuned according to examples, making them very attractive for use in AI. These examples are known as observations or patterns. In supervised learning, each pattern belongs to a certain predefined class. A class can be seen as a decision that has to be made. All the observations combined with their class labels are known as a data set. When a new observation is received, that observation is classified based on previous experience.[203]

A classifier can be trained in various ways; there are many statistical and machine learning approaches. The decision tree[204] is perhaps the most widely used machine learning algorithm. Other widely used classifiers are the neural network,[206]k-nearest neighbor algorithm,[g][208]kernel methods such as the support vector machine (SVM),[h][210]Gaussian mixture model,[211] and the extremely popular naive Bayes classifier.[i][213] Classifier performance depends greatly on the characteristics of the data to be classified, such as the dataset size, the dimensionality, and the level of noise. Model-based classifiers perform well if the assumed model is an extremely good fit for the actual data. Otherwise, if no matching model is available, and if accuracy (rather than speed or scalability) is the sole concern, conventional wisdom is that discriminative classifiers (especially SVM) tend to be more accurate than model-based classifiers such as “naive Bayes” on most practical data sets.[214]

Neural networks, or neural nets, were inspired by the architecture of neurons in the human brain. A simple “neuron” N accepts input from multiple other neurons, each of which, when activated (or “fired”), cast a weighted “vote” for or against whether neuron N should itself activate. Learning requires an algorithm to adjust these weights based on the training data; one simple algorithm (dubbed “fire together, wire together”) is to increase the weight between two connected neurons when the activation of one triggers the successful activation of another. The net forms “concepts” that are distributed among a subnetwork of shared[j] neurons that tend to fire together; a concept meaning “leg” might be coupled with a subnetwork meaning “foot” that includes the sound for “foot”. Neurons have a continuous spectrum of activation; in addition, neurons can process inputs in a nonlinear way rather than weighing straightforward votes. Modern neural nets can learn both continuous functions and, surprisingly, digital logical operations. Neural networks’ early successes included predicting the stock market and (in 1995) a mostly self-driving car.[k] In the 2010s, advances in neural networks using deep learning thrust AI into widespread public consciousness and contributed to an enormous upshift in corporate AI spending; for example, AI-related M&A in 2017 was over 25 times as large as in 2015.[217][218]

The study of non-learning artificial neural networks[206] began in the decade before the field of AI research was founded, in the work of Walter Pitts and Warren McCullouch. Frank Rosenblatt invented the perceptron, a learning network with a single layer, similar to the old concept of linear regression. Early pioneers also include Alexey Grigorevich Ivakhnenko, Teuvo Kohonen, Stephen Grossberg, Kunihiko Fukushima, Christoph von der Malsburg, David Willshaw, Shun-Ichi Amari, Bernard Widrow, John Hopfield, Eduardo R. Caianiello, and others.

The main categories of networks are acyclic or feedforward neural networks (where the signal passes in only one direction) and recurrent neural networks (which allow feedback and short-term memories of previous input events). Among the most popular feedforward networks are perceptrons, multi-layer perceptrons and radial basis networks.[219] Neural networks can be applied to the problem of intelligent control (for robotics) or learning, using such techniques as Hebbian learning (“fire together, wire together”), GMDH or competitive learning.[220]

Today, neural networks are often trained by the backpropagation algorithm, which had been around since 1970 as the reverse mode of automatic differentiation published by Seppo Linnainmaa,[221][222] and was introduced to neural networks by Paul Werbos.[223][224][225]

Hierarchical temporal memory is an approach that models some of the structural and algorithmic properties of the neocortex.[226]

In short, most neural networks use some form of gradient descent on a hand-created neural topology. However, some research groups, such as Uber, argue that simple neuroevolution to mutate new neural network topologies and weights may be competitive with sophisticated gradient descent approaches. One advantage of neuroevolution is that it may be less prone to get caught in “dead ends”.[227]

Deep learning is any artificial neural network that can learn a long chain of causal links. For example, a feedforward network with six hidden layers can learn a seven-link causal chain (six hidden layers + output layer) and has a “credit assignment path” (CAP) depth of seven. Many deep learning systems need to be able to learn chains ten or more causal links in length.[228] Deep learning has transformed many important subfields of artificial intelligence, including computer vision, speech recognition, natural language processing and others.[229][230][228]

According to one overview,[231] the expression “Deep Learning” was introduced to the Machine Learning community by Rina Dechter in 1986[232] and gained traction afterIgor Aizenberg and colleagues introduced it to Artificial Neural Networks in 2000.[233] The first functional Deep Learning networks were published by Alexey Grigorevich Ivakhnenko and V. G. Lapa in 1965.[234][pageneeded] These networks are trained one layer at a time. Ivakhnenko’s 1971 paper[235] describes the learning of a deep feedforward multilayer perceptron with eight layers, already much deeper than many later networks. In 2006, a publication by Geoffrey Hinton and Ruslan Salakhutdinov introduced another way of pre-training many-layered feedforward neural networks (FNNs) one layer at a time, treating each layer in turn as an unsupervised restricted Boltzmann machine, then using supervised backpropagation for fine-tuning. Similar to shallow artificial neural networks, deep neural networks can model complex non-linear relationships. Over the last few years, advances in both machine learning algorithms and computer hardware have led to more efficient methods for training deep neural networks that contain many layers of non-linear hidden units and a very large output layer.[237]

Deep learning often uses convolutional neural networks (CNNs), whose origins can be traced back to the Neocognitron introduced by Kunihiko Fukushima in 1980.[238] In 1989, Yann LeCun and colleagues applied backpropagation to such an architecture. In the early 2000s, in an industrial application CNNs already processed an estimated 10% to 20% of all the checks written in the US.[239]Since 2011, fast implementations of CNNs on GPUs havewon many visual pattern recognition competitions.[228]

CNNs with 12 convolutional layers were used in conjunction with reinforcement learning by Deepmind’s “AlphaGo Lee”, the program that beat a top Go champion in 2016.[240]

Early on, deep learning was also applied to sequence learning with recurrent neural networks (RNNs)[241] which are in theory Turing complete[242] and can run arbitrary programs to process arbitrary sequences of inputs. The depth of an RNN is unlimited and depends on the length of its input sequence; thus, an RNN is an example of deep learning.[228] RNNs can be trained by gradient descent[243][244][245] but suffer from the vanishing gradient problem.[229][246] In 1992, it was shown that unsupervised pre-training of a stack of recurrent neural networks can speed up subsequent supervised learning of deep sequential problems.[247]

Numerous researchers now use variants of a deep learning recurrent NN called the long short-term memory (LSTM) network published by Hochreiter & Schmidhuber in 1997.[248] LSTM is often trained by Connectionist Temporal Classification (CTC).[249] At Google, Microsoft and Baidu this approach has revolutionised speech recognition.[250][251][252] For example, in 2015, Google’s speech recognition experienced a dramatic performance jump of 49% through CTC-trained LSTM, which is now available through Google Voice to billions of smartphone users.[253] Google also used LSTM to improve machine translation,[254] Language Modeling[255] and Multilingual Language Processing.[256] LSTM combined with CNNs also improved automatic image captioning[257] and a plethora of other applications.

AI, like electricity or the steam engine, is a general purpose technology. There is no consensus on how to characterize which tasks AI tends to excel at.[258] While projects such as AlphaZero have succeeded in generating their own knowledge from scratch, many other machine learning projects require large training datasets.[259][260] Researcher Andrew Ng has suggested, as a “highly imperfect rule of thumb”, that “almost anything a typical human can do with less than one second of mental thought, we can probably now or in the near future automate using AI.”[261] Moravec’s paradox suggests that AI lags humans at many tasks that the human brain has specifically evolved to perform well.[125]

Games provide a well-publicized benchmark for assessing rates of progress. AlphaGo around 2016 brought the era of classical board-game benchmarks to a close. Games of imperfect knowledge provide new challenges to AI in the area of game theory.[262][263] E-sports such as StarCraft continue to provide additional public benchmarks.[264][265] There are many competitions and prizes, such as the Imagenet Challenge, to promote research in artificial intelligence. The most common areas of competition include general machine intelligence, conversational behavior, data-mining, robotic cars, and robot soccer as well as conventional games.[266]

The “imitation game” (an interpretation of the 1950 Turing test that assesses whether a computer can imitate a human) is nowadays considered too exploitable to be a meaningful benchmark.[267] A derivative of the Turing test is the Completely Automated Public Turing test to tell Computers and Humans Apart (CAPTCHA). As the name implies, this helps to determine that a user is an actual person and not a computer posing as a human. In contrast to the standard Turing test, CAPTCHA is administered by a machine and targeted to a human as opposed to being administered by a human and targeted to a machine. A computer asks a user to complete a simple test then generates a grade for that test. Computers are unable to solve the problem, so correct solutions are deemed to be the result of a person taking the test. A common type of CAPTCHA is the test that requires the typing of distorted letters, numbers or symbols that appear in an image undecipherable by a computer.

Proposed “universal intelligence” tests aim to compare how well machines, humans, and even non-human animals perform on problem sets that are generic as possible. At an extreme, the test suite can contain every possible problem, weighted by Kolmogorov complexity; unfortunately, these problem sets tend to be dominated by impoverished pattern-matching exercises where a tuned AI can easily exceed human performance levels.[269][270]

AI is relevant to any intellectual task. Modern artificial intelligence techniques are pervasive and are too numerous to list here. Frequently, when a technique reaches mainstream use, it is no longer considered artificial intelligence; this phenomenon is described as the AI effect.

High-profile examples of AI include autonomous vehicles (such as drones and self-driving cars), medical diagnosis, creating art (such as poetry), proving mathematical theorems, playing games (such as Chess or Go), search engines (such as Google search), online assistants (such as Siri), image recognition in photographs, spam filtering, prediction of judicial decisions[273] and targeting online advertisements.[274][275]

With social media sites overtaking TV as a source for news for young people and news organisations increasingly reliant on social media platforms for generating distribution,[276] major publishers now use artificial intelligence (AI) technology to post stories more effectively and generate higher volumes of traffic.[277]

AI is being applied to the high cost problem of dosage issueswhere findings suggested that AI could save $16 billion. In 2016, a ground breaking study in California found that a mathematical formula developed with the help of AI correctly determined the accurate dose of immunosuppressant drugs to give to organ patients.[278]

Artificial intelligence is breaking into the healthcare industry by assisting doctors. According to Bloomberg Technology, Microsoft has developed AI to help doctors find the right treatments for cancer.[279] There is a great amount of research and drugs developed relating to cancer. In detail, there are more than 800 medicines and vaccines to treat cancer. This negatively affects the doctors, because there are too many options to choose from, making it more difficult to choose the right drugs for the patients. Microsoft is working on a project to develop a machine called “Hanover”. Its goal is to memorize all the papers necessary to cancer and help predict which combinations of drugs will be most effective for each patient. One project that is being worked on at the moment is fighting myeloid leukemia, a fatal cancer where the treatment has not improved in decades. Another study was reported to have found that artificial intelligence was as good as trained doctors in identifying skin cancers.[280] Another study is using artificial intelligence to try and monitor multiple high-risk patients, and this is done by asking each patient numerous questions based on data acquired from live doctor to patient interactions.[281]

According to CNN, a recent study by surgeons at the Children’s National Medical Center in Washington successfully demonstrated surgery with an autonomous robot. The team supervised the robot while it performed soft-tissue surgery, stitching together a pig’s bowel during open surgery, and doing so better than a human surgeon, the team claimed.[282] IBM has created its own artificial intelligence computer, the IBM Watson, which has beaten human intelligence (at some levels). Watson not only won at the game show Jeopardy! against former champions,[283] but was declared a hero after successfully diagnosing a woman who was suffering from leukemia.[284]

Advancements in AI have contributed to the growth of the automotive industry through the creation and evolution of self-driving vehicles. As of 2016[update], there are over 30 companies utilizing AI into the creation of driverless cars. A few companies involved with AI include Tesla, Google, and Apple.[285]

Many components contribute to the functioning of self-driving cars. These vehicles incorporate systems such as braking, lane changing, collision prevention, navigation and mapping. Together, these systems, as well as high performance computers, are integrated into one complex vehicle.[286]

Recent developments in autonomous automobiles have made the innovation of self-driving trucks possible, though they are still in the testing phase. The UK government has passed legislation to begin testing of self-driving truck platoons in 2018.[287] Self-driving truck platoons are a fleet of self-driving trucks following the lead of one non-self-driving truck, so the truck platoons aren’t entirely autonomous yet. Meanwhile, the Daimler, a German automobile corporation, is testing the Freightliner Inspiration which is a semi-autonomous truck that will only be used on the highway.[288]

One main factor that influences the ability for a driver-less automobile to function is mapping. In general, the vehicle would be pre-programmed with a map of the area being driven. This map would include data on the approximations of street light and curb heights in order for the vehicle to be aware of its surroundings. However, Google has been working on an algorithm with the purpose of eliminating the need for pre-programmed maps and instead, creating a device that would be able to adjust to a variety of new surroundings.[289] Some self-driving cars are not equipped with steering wheels or brake pedals, so there has also been research focused on creating an algorithm that is capable of maintaining a safe environment for the passengers in the vehicle through awareness of speed and driving conditions.[290]

Another factor that is influencing the ability for a driver-less automobile is the safety of the passenger. To make a driver-less automobile, engineers must program it to handle high-risk situations. These situations could include a head-on collision with pedestrians. The car’s main goal should be to make a decision that would avoid hitting the pedestrians and saving the passengers in the car. But there is a possibility the car would need to make a decision that would put someone in danger. In other words, the car would need to decide to save the pedestrians or the passengers.[291] The programming of the car in these situations is crucial to a successful driver-less automobile.

Financial institutions have long used artificial neural network systems to detect charges or claims outside of the norm, flagging these for human investigation. The use of AI in banking can be traced back to 1987 when Security Pacific National Bank in US set-up a Fraud Prevention Task force to counter the unauthorised use of debit cards. Programs like Kasisto and Moneystream are using AI in financial services.

Banks use artificial intelligence systems today to organize operations, maintain book-keeping, invest in stocks, and manage properties. AI can react to changes overnight or when business is not taking place.[292] In August 2001, robots beat humans in a simulated financial trading competition.[293] AI has also reduced fraud and financial crimes by monitoring behavioral patterns of users for any abnormal changes or anomalies.[294]

The use of AI machines in the market in applications such as online trading and decision making has changed major economic theories.[295] For example, AI based buying and selling platforms have changed the law of supply and demand in that it is now possible to easily estimate individualized demand and supply curves and thus individualized pricing. Furthermore, AI machines reduce information asymmetry in the market and thus making markets more efficient while reducing the volume of trades. Furthermore, AI in the markets limits the consequences of behavior in the markets again making markets more efficient. Other theories where AI has had impact include in rational choice, rational expectations, game theory, Lewis turning point, portfolio optimization and counterfactual thinking.

In video games, artificial intelligence is routinely used to generate dynamic purposeful behavior in non-player characters (NPCs). In addition, well-understood AI techniques are routinely used for pathfinding. Some researchers consider NPC AI in games to be a “solved problem” for most production tasks. Games with more atypical AI include the AI director of Left 4 Dead (2008) and the neuroevolutionary training of platoons in Supreme Commander 2 (2010).[296][297]

Worldwide annual military spending on robotics rose from 5.1 billion USD in 2010 to 7.5 billion USD in 2015.[298][299] Military drones capable of autonomous action are widely considered a useful asset. In 2017, Vladimir Putin stated that “Whoever becomes the leader in (artificial intelligence) will become the ruler of the world”.[300][301] Many artificial intelligence researchers seek to distance themselves from military applications of AI.[302]

For financial statements audit, AI makes continuous audit possible. AI tools could analyze many sets of different information immediately. The potential benefit would be the overall audit risk will be reduced, the level of assurance will be increased and the time duration of audit will be reduced.[303]

A report by the Guardian newspaper in the UK in 2018 found that online gambling companies were using AI to predict the behavior of customers in order to target them with personalized promotions.[304] Developers of commercial AI platforms are also beginning to appeal more directly to casino operators, offering a range of existing and potential services to help them boost their profits and expand their customer base.[305]

Artificial Intelligence has inspired numerous creative applications including its usage to produce visual art. The exhibition “Thinking Machines: Art and Design in the Computer Age, 1959-1989” at MoMA [306] provides a good overview of the historical applications of AI for art, architecture, and design. Recent exhibitions showcasing the usage of AI to produce art include the Google-sponsored benefit and auction at the Gray Area Foundation in San Francisco, where artists experimented with the deepdream algorithm [307] and the exhibition “Unhuman: Art in the Age of AI,” which took place in Los Angeles and Frankfurt in the fall of 2017.[308][309] In the spring of 2018, the Association of Computing Machinery dedicated a special magazine issue to the subject of computers and art highlighting the role of machine learning in the arts.[310]

There are three philosophical questions related to AI:

Go here to read the rest:

Artificial intelligence – Wikipedia

Benefits & Risks of Artificial Intelligence – Future of Life …

Many AI researchers roll their eyes when seeing this headline:Stephen Hawking warns that rise of robots may be disastrous for mankind. And as many havelost count of how many similar articles theyveseen.Typically, these articles are accompanied by an evil-looking robot carrying a weapon, and they suggest we should worry about robots rising up and killing us because theyve become conscious and/or evil.On a lighter note, such articles are actually rather impressive, because they succinctly summarize the scenario that AI researchers dontworry about. That scenario combines as many as three separate misconceptions: concern about consciousness, evil, androbots.

If you drive down the road, you have a subjective experience of colors, sounds, etc. But does a self-driving car have a subjective experience? Does it feel like anything at all to be a self-driving car?Although this mystery of consciousness is interesting in its own right, its irrelevant to AI risk. If you get struck by a driverless car, it makes no difference to you whether it subjectively feels conscious. In the same way, what will affect us humans is what superintelligent AIdoes, not how it subjectively feels.

The fear of machines turning evil is another red herring. The real worry isnt malevolence, but competence. A superintelligent AI is by definition very good at attaining its goals, whatever they may be, so we need to ensure that its goals are aligned with ours. Humans dont generally hate ants, but were more intelligent than they are so if we want to build a hydroelectric dam and theres an anthill there, too bad for the ants. The beneficial-AI movement wants to avoid placing humanity in the position of those ants.

The consciousness misconception is related to the myth that machines cant have goals.Machines can obviously have goals in the narrow sense of exhibiting goal-oriented behavior: the behavior of a heat-seeking missile is most economically explained as a goal to hit a target.If you feel threatened by a machine whose goals are misaligned with yours, then it is precisely its goals in this narrow sense that troubles you, not whether the machine is conscious and experiences a sense of purpose.If that heat-seeking missile were chasing you, you probably wouldnt exclaim: Im not worried, because machines cant have goals!

I sympathize with Rodney Brooks and other robotics pioneers who feel unfairly demonized by scaremongering tabloids,because some journalists seem obsessively fixated on robots and adorn many of their articles with evil-looking metal monsters with red shiny eyes. In fact, the main concern of the beneficial-AI movement isnt with robots but with intelligence itself: specifically, intelligence whose goals are misaligned with ours. To cause us trouble, such misaligned superhuman intelligence needs no robotic body, merely an internet connection this may enable outsmarting financial markets, out-inventing human researchers, out-manipulating human leaders, and developing weapons we cannot even understand. Even if building robots were physically impossible, a super-intelligent and super-wealthy AI could easily pay or manipulate many humans to unwittingly do its bidding.

The robot misconception is related to the myth that machines cant control humans. Intelligence enables control: humans control tigers not because we are stronger, but because we are smarter. This means that if we cede our position as smartest on our planet, its possible that we might also cede control.

See more here:

Benefits & Risks of Artificial Intelligence – Future of Life …

Artificial Intelligence: The Robots Are Now Hiring – WSJ

Sept. 20, 2018 5:30 a.m. ET

Some Fortune 500 companies are using tools that deploy artificial intelligence to weed out job applicants. But is this practice fair? In this episode of Moving Upstream, WSJ’s Jason Bellini investigates.

Some Fortune 500 companies are using tools that deploy artificial intelligence to weed out job applicants. But is this practice fair? In this episode of Moving Upstream, WSJ’s Jason Bellini investigates.

Hiring is undergoing a profound revolution.

Nearly all Fortune 500 companies now use some form of automation — from robot avatars interviewing job candidates to computers weeding out potential employees by scanning keywords in resumes. And more and more companies are using artificial intelligence and machine learning tools to assess possible employees.

DeepSense, based in San Francisco and India, helps hiring managers scan peoples social media accounts to surface underlying personality traits. The company says it uses a scientifically based personality test, and it can be done with or without a potential candidates knowledge.

The practice is part of a general trend of some hiring companies to move away from assessing candidates based on their resumes and skills, towards making hiring decisions based on peoples personalities.

The Robot Revolution: An inside look at how humanoid robots are evolving.

WSJS Jason Bellini explores breakthrough technologies that are reshaping our world and beginning to impact human happiness, health and productivity. Catch the latest episode by signing up here.

Cornell sociology and law professor Ifeoma Ajunwa said shes concerned about these tools potential for bias. Given the large scale of these automatic assessments, she believes potentially faulty algorithms could do more damage than one biased human manager. And she wants scientists to test if the algorithms are fair, transparent and accurate.

In the episode of Moving Upstream above, correspondent Jason Bellini visits South Jordan, Utah-based HireVue, which is delivering AI-based assessments of digital interviews to over 50 companies. HireVue says its algorithm compares candidates tone of voice, word clusters and micro facial expressionsCC with people who have previously been identified as high performers on the job.

Write to Jason Bellini at jason.bellini@wsj.com and Hilke Schellmann at hilke.schellmann@wsj.com

Read the original here:

Artificial Intelligence: The Robots Are Now Hiring – WSJ

Online Artificial Intelligence Courses | Microsoft …

The Microsoft Professional Program (MPP) is a collection of courses that teach skills in several core technology tracks that help you excel in the industry’s newest job roles.

These courses are created and taught by experts and feature quizzes, hands-on labs, and engaging communities. For each track you complete, you earn a certificate of completion from Microsoft proving that you mastered those skills.

Originally posted here:

Online Artificial Intelligence Courses | Microsoft …

What is AI (artificial intelligence)? – Definition from …

AI (artificial intelligence) is the simulation of human intelligence processes by machines, especially computer systems. These processes include learning (the acquisition of information and rules for using the information), reasoning (using rules to reach approximate or definite conclusions) and self-correction. Particular applications of AI include expert systems, speech recognition and machine vision.

AI can be categorized in any number of ways, but here are two examples.

The first classifies AI systems as either weak AI or strong AI.Weak AI, also known as narrow AI, is an AI system that is designed and trained for a particular task. Virtual personal assistants, such as Apple’s Siri, are a form of weak AI.

Strong AI, also known as artificial general intelligence, is an AI system with generalized human cognitive abilities so that when presented with an unfamiliar task, it has enough intelligence to find a solution. TheTuring Test, developed by mathematician Alan Turing in 1950, is a method used to determine if a computer can actually think like a human, although the method is controversial.

Alec Ross on AI and robotics

The second example comes from Arend Hintze, an assistant professor of integrative biology and computer science and engineering at Michigan State University. He categorizes AI into four types, from the kind of AI systems that exist today to sentient systems, which do not yet exist. His categories are as follows:

AI is incorporated into a variety of different types of technology. Here are seven examples.

Artificial intelligence has made its way into a number of areas. Here are six examples.

While AI tools present a range of new functionality for businesses, artificial intellignce also raises some ethical questions. Deep learning algorithms, which underpin many of the most advanced AI tools, only know what’s in the data used during training. Most available data sets for training likely contain traces of human bias. This in turn can make the AI tools biased in their function. This has been seen in the Microsoft chatbot Tay, which learned a misogynistic and anti-Semitic vocabulary from Twitter users, and the Google Photo image classification tool that classified a group of African Americans as gorillas.

The application of AI in the realm of self-driving cars also raises ethical concerns. When an autonomous vehicle is involved in an accident, liability is unclear. Autonomous vehicles may also be put in a position where an accident is unavoidable, forcing it to make ethical decisions about how to minimize damage.

Another major concern is the potential for abuse of AI tools. Hackers are starting to use sophisticated machine learning tools to gain access to sensitive systems, complicating the issue of security beyond its current state.

Deep learning-based video and audio generation tools also present bad actors with the tools necessary to create so-called deepfakes, convincingly fabricated videos of public figures saying or doing things that never took place.

Despite these potential risks, there are few regulations governing the use AI tools, and where laws do exist, the typically pertain to AI only indirectly. For example, federal Fair Lending regulations require financial institutions to explain credit decisions to potential customers, which limit the extent to which lenders can use deep learning algorithms, which by their nature are typically opaque. Europe’s GDPR puts strict limits on how enterprises can use consumer data, which impedes the training and functionality of many consumer-facing AI applications.

In 2016, the National Science and Technology Council issued a report examining the potential role governmental regulation might play in AI development, but it did not recommend specific legislation be considered. Since that time the issue has received little attention from lawmakers.

John McCarthy, an American computer scientist, coined the term “artificial intelligence” in 1956 at the Dartmouth Conference where the discipline was born. Today, it is an umbrella term that encompasses everything from robotic process automation to actual robotics. It has gained prominence recently due, in part, tobig data, or the increase in speed, size and variety of data businesses now collect. AI can perform tasks such as identifying patterns in data more efficiently than humans, enabling businesses to gain more insight from theirdata.

Go here to read the rest:

What is AI (artificial intelligence)? – Definition from …

Artificial Intelligence – Journal – Elsevier

This journal has partnered with Heliyon, an open access journal from Elsevier publishing quality peer reviewed research across all disciplines. Heliyons team of experts provides editorial excellence, fast publication, and high visibility for your paper. Authors can quickly and easily transfer their research from a Partner Journal to Heliyon without the need to edit, reformat or resubmit.>Learn more at Heliyon.com

Here is the original post:

Artificial Intelligence – Journal – Elsevier

What is Artificial Intelligence (AI)? – Definition from …

Artificial intelligence is a branch of computer science that aims to create intelligent machines. It has become an essential part of the technology industry.

Research associated with artificial intelligence is highly technical and specialized. The core problems of artificial intelligence include programming computers for certain traits such as:

Knowledge engineering is a core part of AI research. Machines can often act and react like humans only if they have abundant information relating to the world. Artificial intelligence must have access to objects, categories, properties and relations between all of them to implement knowledge engineering. Initiating common sense, reasoning and problem-solving power in machines is a difficult and tedious task.

Machine learning is also a core part of AI. Learning without any kind of supervision requires an ability to identify patterns in streams of inputs, whereas learning with adequate supervision involves classification and numerical regressions. Classification determines the category an object belongs to and regression deals with obtaining a set of numerical input or output examples, thereby discovering functions enabling the generation of suitable outputs from respective inputs. Mathematical analysis of machine learning algorithms and their performance is a well-defined branch of theoretical computer science often referred to as computational learning theory.

Machine perception deals with the capability to use sensory inputs to deduce the different aspects of the world, while computer vision is the power to analyze visual inputs with a few sub-problems such as facial, object and gesture recognition.

Robotics is also a major field related to AI. Robots require intelligence to handle tasks such as object manipulation and navigation, along with sub-problems of localization, motion planning and mapping.

Original post:

What is Artificial Intelligence (AI)? – Definition from …

19 Artificial Intelligence Technologies That Will Dominate In …

In 2017, we published a popular post on artificial intelligence (AI) technologies that would dominate that year, based on Forresters TechRadar report.

Heres the updated version, which includes 9 more technologies to watch out for this year.

We hope they inspire you to join the 62% of companies boosting their enterprises in 2018.

Natural language generation is an AI sub-discipline that converts data into text, enabling computers to communicate ideas with perfect accuracy.

It is used in customer service to generate reports and market summaries and is offered by companies like Attivio, Automated Insights, Cambridge Semantics, Digital Reasoning, Lucidworks, Narrative Science, SAS, and Yseop.

Siri is just one of the systems that can understand you.

Every day, more and more systems are created that can transcribe human language, reaching hundreds of thousands through voice-response interactive systems and mobile apps.

Companies offering speech recognition services include NICE, Nuance Communications, OpenText and Verint Systems.

A virtual agent is nothing more than a computer agent or program capable of interacting with humans.

The most common example of this kind of technology are chatbots.

Virtual agents are currently being used for customer service and support and as smart home managers.

Some of the companies that provide virtual agents include Amazon, Apple, Artificial Solutions, Assist AI, Creative Virtual, Google, IBM, IPsoft, Microsoft and Satisfi.

These days, computers can also easily learn, and they can be incredibly intelligent!

Machine learning (ML) is a subdiscipline of computer science and a branch of AI. Its goal is to develop techniques that allow computers to learn.

By providing algorithms, APIs (application programming interface), development and training tools, big data, applications and other machines, ML platforms are gaining more and more traction every day.

They are currently mainly being used for prediction and classification.

Some of the companies selling ML platforms include Amazon, Fractal Analytics, Google, H2O.ai, Microsoft, SAS, Skytree and Adext.

This last one is particularly interesting for one simple reason: Adext AI is the first and only audience management tool in the world that applies real AI and machine learning to digital advertising to find the most profitable audience or demographic group for any ad. You can learn more about it here.

AI technology makes hardware much friendlier.

How?

Through new graphic and central processing units and processing devices specifically designed and structured to execute AI-oriented tasks.

And if you havent seen them already, expect the imminent appearance and wide acceptance of AI-optimized silicon chips that can be inserted right into your portable devices and elsewhere.

You can get access to this technology through Alluviate, Cray, Google, IBM, Intel, and Nvidia.

Intelligent machines are capable of introducing rules and logic to AI systems so you can use them for initial setup/training, ongoing maintenance, and tuning.

Decision management has already been incorporated into a variety of corporate applications to assist and execute automated decision, making your business as profitable as possible.

Check out Advanced Systems Concepts, Informatica, Maana, Pegasystems, and UiPath for additional options.

Deep learning platforms use a unique form of ML that involves artificial neural circuits with various abstraction layers that can mimic the human brain, processing data and creating patterns for decision making.

It is currently mainly being used to recognize patterns and classify applications that are only compatible with large-scale data sets.

Deep Instinct, Ersatz Labs, Fluid AI, MathWorks, Peltarion, Saffron Technology and Sentient Technologies all have deep learning options worthy of exploring.

This technology can identify, measure and analyze human behavior and physical aspects of the bodys structure and form.

It allows for more natural interactions between humans and machines, including interactions related to touch, image, speech and body language recognition, and is big within the market research field.

3VR, Affectiva, Agnitio, FaceFirst, Sensory, Synqera and Tahzoo are all biometrics companies working hard to develop this area.

Robotic processes automation uses scripts and methods that mimic and automate human tasks to support corporate processes.

It is particularly useful for situations when hiring humans for a specific job or task is too expensive or inefficient.

The good example is Adext AI, a platform that automates digital advertising processes using AI, saving businesses from devoting hours to mechanical and repetitive tasks.

Its a solution that lets you make the most of your human talent and move employees into more strategic and creative positions, so their actions can really make an impact on the company’s growth.

Advanced Systems Concepts, Automation Anywhere, Blue Prism, UiPath, and WorkFusion are other examples of robotic processes automation companies.

This technology uses text analytics to understand the structure of sentences, as well as their meaning and intention, through statistical methods and ML.

Text analytics and NLP are currently being used for security systems and fraud detection.

They are also being used by a vast array of automated assistants and apps to extract unstructured data.

Some of the service providers and suppliers of these technologies include Basis Technology, Coveo, Expert System, Indico, Knime, Lexalytics, Linguamatics, Mindbreeze, Sinequa, Stratifyd, and Synapsify.

A digital twin is a software construct that bridges the gap between physical systems and the digital world.

General Electric (GE), for example, is building an AI workforce to monitor its aircraft engines, locomotives and gas turbines and predict failures with cloud-hosted software models of GEs machines. Their digital twins are mainly lines of software code, but the most elaborate versions look like 3-D computer-aided design drawings full of interactive charts, diagrams, and data points.

Companies using digital twin and AI modeling technologies include VEERUM, in the capital project delivery space; Akselos, which is using it to protect critical infrastructure, and Supply Dynamics, which has developed a SaaS solution to manage raw material sourcing in complex, highly distributed manufacturing environments.

Cyber defense is a computer network defense mechanism that focuses on preventing, detecting and providing timely responses to attacks or threats to infrastructure and information.

AI and ML are now being used to move cyberdefense into a new evolutionary phase in response to an increasingly hostile environment: Breach Level Index detected a total of over 2 billion breached records during 2017. Seventy-six percent of the records in the survey were lost accidentally, and 69% were an identity theft type of breach.

Recurrent neural networks, which are capable of processing sequences of inputs, can be used in combination with ML techniques to create supervised learning technologies, which uncover suspicious user activity and detect up to 85% of all cyber attacks.

Startups such as Darktrace, which pairs behavioral analytics with advanced mathematics to automatically detect abnormal behavior within organizations and Cylance, which applies AI algorithms to stop malware and mitigate damage from zero-day attacks, are both working in the area of AI-powered cyber defense.

DeepInstinct, another cyber defense company, is a deep learning project named Most Disruptive Startup by Nvidias Silicon Valley ceremony, protects enterprises’ endpoints, servers, and mobile devices.

Compliance is the certification or confirmation that a person or organization meets the requirements of accepted practices, legislation, rules and regulations, standards or the terms of a contract, and there is a significant industry that upholds it.

We are now seeing the first wave of regulatory compliance solutions that use AI to deliver efficiency through automation and comprehensive risk coverage.

Some examples of AIs use in compliance are showing up across the world. For example, NLP (Natural Language Processing) solutions can scan regulatory text and match its patterns with a cluster of keywords to identify the changes that are relevant to an organization.

Capital stress testing solutions with predictive analytics and scenario builders can help organizations stay compliant with regulatory capital requirements. And the volume of transaction activities flagged as potential examples of money laundering can be reduced as deep learning is used to apply increasingly sophisticated business rules to each one.

Companies working in this area include Compliance.ai, a Retch company that matches regulatory documents to a corresponding business function; Merlon Intelligence, a global compliance technology company that supports the financial services industry to combat financial crimes, and Socure, whose patented predictive analytics platform boosts customer acceptance rates while reducing fraud and manual reviews.

While some are rightfully concerned about AI replacing people in the workplace, lets not forget that AI technology also has the potential to vastly help employees in their work, especially those in knowledge work.

In fact, the automation of knowledge work has been listed as the #2 most disruptive emerging tech trend.

The medical and legal professions, which are heavily reliant on knowledge workers, is where workers will increasingly use AI as a diagnostic tool.

There is an increasing number of companies working on technologies in this area. Kim Technologies, whose aim is to empower knowledge workers who have little to no IT programming experience with the tools to create new workflow and document processes with the help of AI, is one of them. Kyndi is another, whose platform is designed to help knowledge workers process vast amounts of information.

Content creation now includes any material people contribute to the online world, such as videos, ads, blog posts, white papers, infographics and other visual or written assets.

Brands like USA Today, Hearst and CBS, are already using AI to generate their content.

Wibbitz, a SaaS tool that helps publishers create videos from written content in minutes with AI video production technology, is a great example of a solution from this field. Wordsmith is another tool, created by Automated Insights, that applies NLP (Natural Language Processing) to generate news stories based on earnings data.

Peer-to-peer networks, in their purest form, are created when two or more PCs connect and share resources without the data going through a server computer.

But peer-to-peer networks are also used by cryptocurrencies, and have the potential to even solve some of the worlds most challenging problems, by collecting and analyzing large amounts of data, says Ben Hartman, CEO of Bet Capital LLC, to Entrepreneur.

Nano Vision, a startup that rewards users with cryptocurrency for their molecular data, aims to change the way we approach threats to human health, such as superbugs, infectious diseases, and cancer, among others.

Another player utilizing peer-to-peer networks and AI is Presearch, a decentralized search engine thats powered by the community and rewards members with tokens for a more transparent search system.

This technology allows software to read the emotions on a human face using advanced image processing or audio data processing. We are now at the point where we can capture micro-expressions, or subtle body language cues, and vocal intonation that betrays a persons feelings.

Law enforcers can use this technology to try to detect more information about someone during interrogation. But it also has a wide range of applications for marketers.

There are increasing numbers of startups working in this area. Beyond Verbal analyzes audio inputs to describe a persons character traits, including how positive, how excited, angry or moody they are. nViso uses emotion video analytics to inspire new product ideas, identify upgrades and enhance the consumer experience. And Affectivas Emotion AI is used in the gaming, automotive, robotics, education, healthcare industries, and other fields, to apply facial coding and emotion analytics from face and voice data.

Image recognition is the process of identifying and detecting an object or feature in a digital image or video, and AI is increasingly being stacked on top of this technology to great effect.

AI can search social media platforms for photos and compare them to a wide range of data sets to decide which ones are most relevant during image searches.

Image recognition technology can also be used to detect license plates, diagnose disease, analyze clients and their opinions and verify users based on their face.

Clarifai provides image recognition systems for customers to detect near-duplicates and find similar uncategorized images.

SenseTime is one of the leaders in this industry and develops face recognition technology that can be applied to payment and picture analysis for bank card verification and other applications. And GumGums mission is to unlock the value of images and videos produced across the web using AI technology.

View post:

19 Artificial Intelligence Technologies That Will Dominate In …

Artificial Intelligence (AI) and Machine Learning | Oracle

Complete AI Portfolio

Oracle offers a complete portfolio of products, services, and differentiated capabilities to power your enterprise with artificial intelligence. For business users, Oracle offers ready-to-go AI- powered cloud applications with intelligent features that drive better business outcomes. With Oracles ready-to-build AI Platform, data scientists and application developers have a full suite of cloud services to build, deploy, and manage AI-powered solutions. With Oracles ready-to-work Autonomous Database, machine learning is working behind the scenes to automate security patching, backups, and optimize database query performance, which helps to eliminate human error and repetitive manual tasks so organizations can focus on higher-value activities.

See the original post here:

Artificial Intelligence (AI) and Machine Learning | Oracle

What is Artificial Intelligence? AI Basics Explained | StopAd

When someone mentions artificial intelligence(AI), what is the first thing that comes to your mind?

Most of us imagine an army of human-like robots rebelling against humanity, while a fewfolks from a more positive thinking communityare envisioning a bright future where AI serves them in all possible ways from walking a dog early in the morning when the weather is unpleasant to peeling tangerines at Christmas.

While the second scenario is certainly fun, it is a utopia per se. A reality with robots rising up against people, however, is a much more probable event.

If you believe AI will soon become the greatest existential threat to humanity, weve got some good news for you. Elon Musk and Stephen Hawking share your point of view. (What a way to boost your self-esteem, right?)

Jokes aside, the debate within the global tech community is not centered on the impact of human-like AIas the general public thinksbut rather on the possibility of ever achieving this technology outright. Professionals are absorbed in discussions about how to define human-like and intelligence. These definitions may seem trivial to outsiders but understanding the human mind and intelligence are, in fact, critical to determining the timeline of milestones for AI. Experts are still not certain how this kind of intelligence will manifest or how soon day X will come, but it is clear that we are moving towards this reality with increasing speed.

This means it is high time to finally understand what AI is all about.

First things first. Before digging deeper into the topic of AI, lets briefly discuss what artificial intelligence is and how it works.

The term artificial intelligence dates back to 1956 and belongs to a Stanford researcher John McCarthy, who coined the term and defined the key mission of AI as a sub-field of computer science.

Basically, artificial intelligence (AI) is the ability of a machine or a computer program to think and learn. The concept of AI is based on the idea of building machines capable of thinking, acting, and learning like humans.

A more nuanced definition is that artificial Intelligence is an interdisciplinary concept that studies the possibility of creating machines capable of interacting with their environment and acting upon the received data in the a manner considered intelligent.

While some people falsely consider AI a technology, the more accurate approach would be seeing it as a broad concept in which machines are able to deal with tasks in a way we would call intelligent or smart.

There are certain things a machine/computer program must be capable of to be considered AI.

First, it should be able to mimic human thought process and behavior. Second, it should act in a human-like wayintelligent, rational, and ethical.

It is worth mentioning that the AI concept relates both to Weak AI and General AI that has cognitive functions. Stanford has outlined a helpful AI FAQ on these topics.

Not really. Although the two terms are often used interchangeably, they are not the same.

Artificial intelligence is a broader concept, while machine learning is the most common application of AI.

We should understand machine learning as a current application of AI that is focused on development of computer programs that can access data and learn from it automatically, without human assistance or intervention. The entire machine learning concept is based on the assumption that we should give machines access to information and let them learn from it themselves.

Artificial intelligence, in its turn, is a bunch of technologies that include machine learning and some other technologies like natural language processing, inference algorithms, neural networks, etc.

Many people associate AI with the distant future. They incorrectly believe that despite all the buzz around artificial intelligence, the technology is not likely to become a part of their lives anytime soon. Little do they know how many aspects of their lives are already affected by AI.

Take Siri or Alexapersonal assistants that have already become the new normal for thousands of people around the globe. These and similar intelligent gadgets are able to recognize our speech (read: understand what we want or need), analyze the information they have access to, and provide an answer or solution. What is remarkable (and a little scary) about such assistants is that they continuously learn about their users until the point at which they are able to accurately anticipate users needs.

Spotify, Pandora, and Apple Music are some other touching points between AI and you. These services are capable of recommending music based on your interests. These apps monitor the choices you make, insert them into a learning algorithm, and suggest music you are most likely to enjoy. This particular use of AI is probably one of the simplest among all, but it does a good job helping us discover new songs and artists.

AI is making headway in areas you might least expect it. For example, when you come across short news stories on the Associated Press or Yahoo!, chances are good they were written by AI. The current state of artificial intelligence already allows for some basic robot writing. It might be not yet ready to compose in-depth articles or creative stories, but does a pretty good job writing short and simple articles like sport recaps and financial summaries.

Other examples of artificial intelligence in use today include smart home devices like Googles NEST, self-driving cars like those produced by Tesla, and online games like Alien: Isolation.

Here at StopAd, we rely on artificial intelligence, too.

Thanks to the AI weve developed, our ad blocker is able to detect ads just like a human does. This means identifying and blocking ads regardless of their placement, size, type, and format. StopAd is even capable of identifying native advertisingads designed to mimic the structure and layout of the website they appear on. Furthermore, we sometimes use AI to conduct our own investigations.

Some people claim that AI is still in its infancy. Others assure us that we are only a few years away from AI gaining control over humanity. The truth, however, lies somewhere in between.

According to the most trustworthy forecasts out there, AI will outsmart humans at virtually everything in the following 45 years. Obviously, this wont happen overnight. Industries will be falling under AIs spell one-by-one.

Experts predict that within the next decade AI will outperform humans in relatively simple tasks such as translating languages, writing school essays, and driving trucks. More complicated tasks like writing a bestselling book or working as a surgeon, however, will take machines much more time to learn. AI is expected to master these two skills by 2049 and 2053 accordingly.

It is obviously too soon to talk about AI-powered creatures like those from Westworld or Ex Machina stealing our jobs or, worse yet, rising against humanity, but we are certainly moving in that direction. Meanwhile, top tech professionals and scientists are getting increasingly concerned about our future and encourage further research on the potential impact of AI.

It looks like those who understand the full potential of AI are more scared of it than those who only know the basics. A recent scandal between Googles executives and employees may serve as a proof. In April, employees of Google demanded the company to stop working on a so-called Pentagon Project as they were afraid of being involved in the business of war. The project officially known as Project Maven is meant to use AI to make it easier to classify images of people and objects shot by drones. The potential danger is that the life-or-death decisions of what needs to be bombarded and what doesnt will be made without humans involvement.

The military explains that their only intent is to reduce the current workload and minimize the number of tedious tasks performed by humanssomething AI is extremely well-suited for.

Given that lives of people might be at stake, however, can these tasks even be called tedious? And theres another critical question. In a world like this, who will bear the blame of killing innocent people?

It is a widespread point of view that one day not only will AI exceed human performance but it will also extend beyond human control. With so many fearful articles out there, questions like is artificial intelligence safe? or is artificial intelligence bad for people? should come as no surprise. AI is obviously exciting but simultaneously warrants caution.

Given the innate advantage AI machines have over us humans (accuracy, speed, etc.) an AI rebellion scenario is something we should not completely dismiss. Time will show us whether AI is our greatest existential threat or a tech blessing that will improve our quality of life in many different ways.

So far, one thing remains perfectly clear: creating AI is one of the most remarkable events for humankind. After all, AI is considered a major component of 4th Industrial Revolution, and its potential socioeconomic impact is believed to be as huge as the invention of electricity once had.

In light of this, the smartest approach would be keeping an eye on how the technology evolves, taking advantage of the improvements it brings to our lives, and not getting too nervous at the thought of machine takeover.

StopAd is the most effective and easy-to-use ad blocker on the market. Powered by artificial intelligence, StopAd detects ads nearly as well as a human and blocks them on all browsers without multiple downloads. Install StopAd to enjoy an ad-free online experience.

Learn more

Follow this link:

What is Artificial Intelligence? AI Basics Explained | StopAd

B.S. in Artificial Intelligence | Carnegie Mellon School of …

Carnegie Mellon has led the world in artificial intelligence education and innovation since the field was created. It’s only natural, then, that the School of Computer Science would offer the nation’s first bachelor’s degree in artificial intelligence, which we introduced in fall 2018.

The BSAI program gives you the in-depth knowledge you need to transform large amounts of data into actionable decisions. The program and its curriculum focus on how complex inputs such as vision, language and huge databases can be used to make decisions or enhance human capabilities. The curriculum includes coursework in computer science, math, statistics, computational modeling, machine learning and symbolic computation. Because CMU is devoted to AI for social good, you’ll also take courses in ethics and social responsibility, with the option to participate in independent study projects that change the world for the better in areas like healthcare, transportation and education.

Just as AI unites disciplines from machine learning to natural language processing, instruction in the BSAI program includes faculty members from the school’s Computer Science Department, Human-Computer Interaction Institute, Institute for Software Research, Language Technologies Institute, Machine Learning Department and Robotics Institute.

When you graduate with a B.S. in AI from SCS, you’ll have the computer science savvy and skills our students are known for, with the added expertise in machine learning and automated reasoning that you’ll need to build the AI of tomorrow.

See the Curriculum

The BSAI program is reserved for current and future SCS students only, so you need to be accepted into the School of Computer Science first. Once you’re at Carnegie Mellon and enrolled in SCS, you can declare a BSAI major in the spring of your first year. Initially, the program will accommodate roughly 100 students total, or about 3035 from each class.

Learn More About Admissions

If you’re an SCS student interested in applying for the BSAI program, send us an email.

See the original post:

B.S. in Artificial Intelligence | Carnegie Mellon School of …

Artificial Intelligence research at Microsoft

At Microsoft, researchers in artificial intelligence are harnessing the explosion of digital data and computational power with advanced algorithms to enable collaborative and natural interactions between people and machines that extend the human ability to sense, learn and understand. The research infuses computers, materials and systems with the ability to reason, communicate and perform with humanlike skill and agility.

Microsofts deep investments in the field are advancing the state of the art in machine intelligence and perception, enabling computers that understand what they see, communicate in natural language, answer complex questions and interact with their environment. In addition, the companys researchers are thought leaders on the ethics and societal impacts of intelligent technologies. The research, tools and services that result from this investment are woven into existing and new products and, at the same time, made open and accessible to the broader community in a bid to accelerate innovation, democratize AI and solve the worlds most pressing challenges.

Here is the original post:

Artificial Intelligence research at Microsoft

A.I. Artificial Intelligence – Wikipedia

A.I. Artificial Intelligence, also known as A.I., is a 2001 American science fiction drama film co-written, co-produced, and directed by Steven Spielberg, co-written by Ian Watson, and based on the 1969 short story “Supertoys Last All Summer Long” by Brian Aldiss. It was co-produced by Kathleen Kennedy and Bonnie Curtis, and stars Haley Joel Osment, Jude Law, Frances O’Connor, Brendan Gleeson, and William Hurt. Set in a futuristic post-climate change society, A.I. tells the story of David (Osment), a childlike android uniquely programmed with the ability to love.

Development of A.I. originally began with producer-director Stanley Kubrick, after he acquired the rights to Aldiss’ story in the early 1970s. Kubrick hired a series of writers until the mid-1990s, including Brian Aldiss, Bob Shaw, Ian Watson, and Sara Maitland. The film languished in protracted development for years, partly because Kubrick felt computer-generated imagery was not advanced enough to create the David character, whom he believed no child actor would convincingly portray. In 1995, Kubrick handed A.I. to Spielberg, but the film did not gain momentum until Kubrick’s death in 1999. Spielberg remained close to Watson’s film treatment for the screenplay.

A.I. Artificial Intelligence divided critics, with the overall balance being positive, and grossed approximately $235 million. It was nominated for two Academy Awards at the 74th Academy Awards for Best Visual Effects and Best Original Score (by John Williams). In a 2016 BBC poll of 177 critics around the world, the film was voted the eighty-third greatest film since 2000.[3]

In the late 22nd century, rising sea levels from global warming have wiped out coastal cities such as Amsterdam, Venice, and New York and drastically reduced the world’s population. A new type of robots called Mecha, advanced humanoids capable of thought and emotion, have been created.

David, a Mecha that resembles a human child and is programmed to display love for his owners, is given to Henry Swinton and his wife Monica, whose son Martin, after contracting a rare disease, has been placed in suspended animation and not expected to recover. Monica feels uneasy with David, but eventually warms to him and activates his imprinting protocol, causing him to have an enduring childlike love for her. David is befriended by Teddy, a robotic teddy bear that belonged to Martin.

Martin is cured of his disease and brought home. As he recovers, he grows jealous of David. He tricks David into entering the parents’s bedroom at night and cutting off a lock of Monica’s hair. This upsets the parents, particularly Henry, who fears David intended to injure them.

At a pool party, one of Martin’s friends pokes David with a knife, activating David’s self-protection programming. David grabs Martin and they fall into the pool. Martin is saved from drowning, but Henry persuades Monica to return David to his creators for destruction. Instead, she abandons David and Teddy in the forest. She warns David to avoid all humans, and tells him to find other unregistered Mecha who can protect him.

David is captured for an anti-Mecha “Flesh Fair”, where obsolete, unlicensed Mecha are destroyed before cheering crowds. David is placed on a platform with Gigolo Joe, a male prostitute Mecha who is on the run after being framed for murder. Before the pair can be destroyed with acid, the crowd, thinking David is a real boy, begins booing and throwing things at the show’s emcee. In the chaos, David and Joe escape. They set out to find the Blue Fairy, whom David remembers from The Adventures of Pinocchio, and who David believes can turn him into a real boy, allowing Monica to love him and take him home.

Joe and David make their way to the decadent resort town of Rouge City, where “Dr. Know”, a holographic answer engine, directs them to the top of Rockefeller Center in the flooded ruins of Manhattan. There, David meets a copy of himself and destroys it. He then meets Professor Hobby, his creator, who tells David that he was built in the image of the professor’s dead son David. David finds more copies of him, as well as female versions called Darlene, that have been made there.

Disheartened, David lets himself fall from a ledge of the building. He is rescued by Joe, flying an amphibicopter he has stolen from the police who were pursuing him. David tells Joe he saw the Blue Fairy underwater, and wants to go down to meet her. Joe is captured by the authorities, who snare him with an electromagnet. David and Teddy use the amphibicopter to dive down to see the Fairy, which turns out to be a statue at the now-sunken Coney Island. The two become trapped when the Wonder Wheel falls on their vehicle. David repeatedly asks the Fairy to turn him into a real boy. Eventually the ocean freezes and David’s power source is depleted.

Two thousand years later, humans are extinct, and Manhattan is buried under glacial ice. The Mecha have evolved into an advanced silicon-based form called Specialists. They find David and Teddy, and discover they are original Mecha who knew living humans, making them special.

The Specialists revive David and Teddy. David walks to the frozen Fairy statue, which collapses when he touches it. The Mecha use David’s memories to reconstruct the Swinton home. David asks the Specialists if they can make him human, but they cannot. However, he insists they recreate Monica from DNA from the lock of her hair, which Teddy has kept. The Mecha warn David that the clone can live for only a day, and that the process cannot be repeated. David spends the next day with Monica and Teddy. Before she drifts off to sleep, Monica tells David she has always loved him. Teddy climbs onto the bed and watches the two lie peacefully together.

Kubrick began development on an adaptation of “Super-Toys Last All Summer Long” in the late 1970s, hiring the story’s author, Brian Aldiss, to write a film treatment. In 1985, Kubrick asked Steven Spielberg to direct the film, with Kubrick producing.[6] Warner Bros. agreed to co-finance A.I. and cover distribution duties.[7] The film labored in development hell, and Aldiss was fired by Kubrick over creative differences in 1989.[8] Bob Shaw briefly served as writer, leaving after six weeks due to Kubrick’s demanding work schedule, and Ian Watson was hired as the new writer in March 1990. Aldiss later remarked, “Not only did the bastard fire me, he hired my enemy [Watson] instead.” Kubrick handed Watson The Adventures of Pinocchio for inspiration, calling A.I. “a picaresque robot version of Pinocchio”.[7][9]

Three weeks later, Watson gave Kubrick his first story treatment, and concluded his work on A.I. in May 1991 with another treatment of 90 pages. Gigolo Joe was originally conceived as a G.I. Mecha, but Watson suggested changing him to a male prostitute. Kubrick joked, “I guess we lost the kiddie market.”[7] Meanwhile, Kubrick dropped A.I. to work on a film adaptation of Wartime Lies, feeling computer animation was not advanced enough to create the David character. However, after the release of Spielberg’s Jurassic Park, with its innovative computer-generated imagery, it was announced in November 1993 that production of A.I. would begin in 1994.[10] Dennis Muren and Ned Gorman, who worked on Jurassic Park, became visual effects supervisors,[8] but Kubrick was displeased with their previsualization, and with the expense of hiring Industrial Light & Magic.[11]

“Stanley [Kubrick] showed Steven [Spielberg] 650 drawings which he had, and the script and the story, everything. Stanley said, ‘Look, why don’t you direct it and I’ll produce it.’ Steven was almost in shock.”

Producer Jan Harlan, on Spielberg’s first meeting with Kubrick about A.I.[12]

In early 1994, the film was in pre-production with Christopher “Fangorn” Baker as concept artist, and Sara Maitland assisting on the story, which gave it “a feminist fairy-tale focus”.[7] Maitland said that Kubrick never referred to the film as A.I., but as Pinocchio.[11] Chris Cunningham became the new visual effects supervisor. Some of his unproduced work for A.I. can be seen on the DVD, The Work of Director Chris Cunningham.[13] Aside from considering computer animation, Kubrick also had Joseph Mazzello do a screen test for the lead role.[11] Cunningham helped assemble a series of “little robot-type humans” for the David character. “We tried to construct a little boy with a movable rubber face to see whether we could make it look appealing,” producer Jan Harlan reflected. “But it was a total failure, it looked awful.” Hans Moravec was brought in as a technical consultant.[11]Meanwhile, Kubrick and Harlan thought A.I. would be closer to Steven Spielberg’s sensibilities as director.[14][15] Kubrick handed the position to Spielberg in 1995, but Spielberg chose to direct other projects, and convinced Kubrick to remain as director.[12][16] The film was put on hold due to Kubrick’s commitment to Eyes Wide Shut (1999).[17] After the filmmaker’s death in March 1999, Harlan and Christiane Kubrick approached Spielberg to take over the director’s position.[18][19] By November 1999, Spielberg was writing the screenplay based on Watson’s 90-page story treatment. It was his first solo screenplay credit since Close Encounters of the Third Kind (1977).[20] Spielberg remained close to Watson’s treatment, but removed various sex scenes with Gigolo Joe. Pre-production was briefly halted during February 2000, because Spielberg pondered directing other projects, which were Harry Potter and the Philosopher’s Stone, Minority Report and Memoirs of a Geisha.[17][21] The following month Spielberg announced that A.I. would be his next project, with Minority Report as a follow-up.[22] When he decided to fast track A.I., Spielberg brought Chris Baker back as concept artist.[16]

The original start date was July 10, 2000,[15] but filming was delayed until August.[23] Aside from a couple of weeks shooting on location in Oxbow Regional Park in Oregon, A.I. was shot entirely using sound stages at Warner Bros. Studios and the Spruce Goose Dome in Long Beach, California.[24]The Swinton house was constructed on Stage 16, while Stage 20 was used for Rouge City and other sets.[25][26] Spielberg copied Kubrick’s obsessively secretive approach to filmmaking by refusing to give the complete script to cast and crew, banning press from the set, and making actors sign confidentiality agreements. Social robotics expert Cynthia Breazeal served as technical consultant during production.[15][27] Haley Joel Osment and Jude Law applied prosthetic makeup daily in an attempt to look shinier and robotic.[4] Costume designer Bob Ringwood (Batman, Troy) studied pedestrians on the Las Vegas Strip for his influence on the Rouge City extras.[28] Spielberg found post-production on A.I. difficult because he was simultaneously preparing to shoot Minority Report.[29]

The film’s soundtrack was released by Warner Sunset Records in 2001. The original score was composed and conducted by John Williams and featured singers Lara Fabian on two songs and Josh Groban on one. The film’s score also had a limited release as an official “For your consideration Academy Promo”, as well as a complete score issue by La-La Land Records in 2015.[30] The band Ministry appears in the film playing the song “What About Us?” (but the song does not appear on the official soundtrack album).

Warner Bros. used an alternate reality game titled The Beast to promote the film. Over forty websites were created by Atomic Pictures in New York City (kept online at Cloudmakers.org) including the website for Cybertronics Corp. There were to be a series of video games for the Xbox video game console that followed the storyline of The Beast, but they went undeveloped. To avoid audiences mistaking A.I. for a family film, no action figures were created, although Hasbro released a talking Teddy following the film’s release in June 2001.[15]

A.I. had its premiere at the Venice Film Festival in 2001.[31]

A.I. Artificial Intelligence was released on VHS and DVD by Warner Home Video on March 5, 2002 in both a standard full-screen release with no bonus features, and as a 2-Disc Special Edition featuring the film in its original 1.85:1 anamorphic widescreen format as well as an eight-part documentary detailing the film’s development, production, music and visual effects. The bonus features also included interviews with Haley Joel Osment, Jude Law, Frances O’Connor, Steven Spielberg and John Williams, two teaser trailers for the film’s original theatrical release and an extensive photo gallery featuring production sills and Stanley Kubrick’s original storyboards.[32]

The film was released on Blu-ray Disc on April 5, 2011 by Paramount Home Media Distribution for the U.S. and by Warner Home Video for international markets. This release featured the film a newly restored high-definition print and incorporated all the bonus features previously included on the 2-Disc Special Edition DVD.[33]

The film opened in 3,242 theaters in the United States on June 29, 2001, earning $29,352,630 during its opening weekend. A.I went on to gross $78.62 million in US totals as well as $157.31 million in foreign countries, coming to a worldwide total of $235.93 million.[34]

Based on 192 reviews collected by Rotten Tomatoes, 73% of critics gave the film positive notices with a score of 6.6/10. The website’s critical consensus reads, “A curious, not always seamless, amalgamation of Kubrick’s chilly bleakness and Spielberg’s warm-hearted optimism. A.I. is, in a word, fascinating.”[35] By comparison, Metacritic collected an average score of 65, based on 32 reviews, which is considered favorable.[36]

Producer Jan Harlan stated that Kubrick “would have applauded” the final film, while Kubrick’s widow Christiane also enjoyed A.I.[37] Brian Aldiss admired the film as well: “I thought what an inventive, intriguing, ingenious, involving film this was. There are flaws in it and I suppose I might have a personal quibble but it’s so long since I wrote it.” Of the film’s ending, he wondered how it might have been had Kubrick directed the film: “That is one of the ‘ifs’ of film historyat least the ending indicates Spielberg adding some sugar to Kubrick’s wine. The actual ending is overly sympathetic and moreover rather overtly engineered by a plot device that does not really bear credence. But it’s a brilliant piece of film and of course it’s a phenomenon because it contains the energies and talents of two brilliant filmmakers.”[38] Richard Corliss heavily praised Spielberg’s direction, as well as the cast and visual effects.[39] Roger Ebert gave the film three stars, saying that it was “wonderful and maddening.”[40] Leonard Maltin, on the other hand, gives the film two stars out of four in his Movie Guide, writing: “[The] intriguing story draws us in, thanks in part to Osment’s exceptional performance, but takes several wrong turns; ultimately, it just doesn’t work. Spielberg rewrote the adaptation Stanley Kubrick commissioned of the Brian Aldiss short story ‘Super Toys Last All Summer Long’; [the] result is a curious and uncomfortable hybrid of Kubrick and Spielberg sensibilities.” However, he calls John Williams’ music score “striking”. Jonathan Rosenbaum compared A.I. to Solaris (1972), and praised both “Kubrick for proposing that Spielberg direct the project and Spielberg for doing his utmost to respect Kubrick’s intentions while making it a profoundly personal work.”[41] Film critic Armond White, of the New York Press, praised the film noting that “each part of Davids journey through carnal and sexual universes into the final eschatological devastation becomes as profoundly philosophical and contemplative as anything by cinemas most thoughtful, speculative artists Borzage, Ozu, Demy, Tarkovsky.”[42] Filmmaker Billy Wilder hailed A.I. as “the most underrated film of the past few years.”[43] When British filmmaker Ken Russell saw the film, he wept during the ending.[44]

Mick LaSalle gave a largely negative review. “A.I. exhibits all its creators’ bad traits and none of the good. So we end up with the structureless, meandering, slow-motion endlessness of Kubrick combined with the fuzzy, cuddly mindlessness of Spielberg.” Dubbing it Spielberg’s “first boring movie”, LaSalle also believed the robots at the end of the film were aliens, and compared Gigolo Joe to the “useless” Jar Jar Binks, yet praised Robin Williams for his portrayal of a futuristic Albert Einstein.[45][not in citation given] Peter Travers gave a mixed review, concluding “Spielberg cannot live up to Kubrick’s darker side of the future.” But he still put the film on his top ten list that year for best movies.[46] David Denby in The New Yorker criticized A.I. for not adhering closely to his concept of the Pinocchio character. Spielberg responded to some of the criticisms of the film, stating that many of the “so called sentimental” elements of A.I., including the ending, were in fact Kubrick’s and the darker elements were his own.[47] However, Sara Maitland, who worked on the project with Kubrick in the 1990s, claimed that one of the reasons Kubrick never started production on A.I. was because he had a hard time making the ending work.[48] James Berardinelli found the film “consistently involving, with moments of near-brilliance, but far from a masterpiece. In fact, as the long-awaited ‘collaboration’ of Kubrick and Spielberg, it ranks as something of a disappointment.” Of the film’s highly debated finale, he claimed, “There is no doubt that the concluding 30 minutes are all Spielberg; the outstanding question is where Kubrick’s vision left off and Spielberg’s began.”[49]

Screenwriter Ian Watson has speculated, “Worldwide, A.I. was very successful (and the 4th highest earner of the year) but it didn’t do quite so well in America, because the film, so I’m told, was too poetical and intellectual in general for American tastes. Plus, quite a few critics in America misunderstood the film, thinking for instance that the Giacometti-style beings in the final 20 minutes were aliens (whereas they were robots of the future who had evolved themselves from the robots in the earlier part of the film) and also thinking that the final 20 minutes were a sentimental addition by Spielberg, whereas those scenes were exactly what I wrote for Stanley and exactly what he wanted, filmed faithfully by Spielberg.”[50]

In 2002, Spielberg told film critic Joe Leydon that “People pretend to think they know Stanley Kubrick, and think they know me, when most of them don’t know either of us”. “And what’s really funny about that is, all the parts of A.I. that people assume were Stanley’s were mine. And all the parts of A.I. that people accuse me of sweetening and softening and sentimentalizing were all Stanley’s. The teddy bear was Stanley’s. The whole last 20 minutes of the movie was completely Stanley’s. The whole first 35, 40 minutes of the film all the stuff in the house was word for word, from Stanley’s screenplay. This was Stanley’s vision.” “Eighty percent of the critics got it all mixed up. But I could see why. Because, obviously, I’ve done a lot of movies where people have cried and have been sentimental. And I’ve been accused of sentimentalizing hard-core material. But in fact it was Stanley who did the sweetest parts of A.I., not me. I’m the guy who did the dark center of the movie, with the Flesh Fair and everything else. That’s why he wanted me to make the movie in the first place. He said, ‘This is much closer to your sensibilities than my own.'”[51]

Upon rewatching the film many years after its release, BBC film critic Mark Kermode apologized to Spielberg in an interview in January 2013 for “getting it wrong” on the film when he first viewed it in 2001. He now believes the film to be Spielberg’s “enduring masterpiece”.[52]

Visual effects supervisors Dennis Muren, Stan Winston, Michael Lantieri and Scott Farrar were nominated for the Academy Award for Best Visual Effects, while John Williams was nominated for Best Original Music Score.[53] Steven Spielberg, Jude Law and Williams received nominations at the 59th Golden Globe Awards.[54] A.I. was successful at the Saturn Awards, winning five awards, including Best Science Fiction Film along with Best Writing for Spielberg and Best Performance by a Younger Actor for Osment.[55]

Read more from the original source:

A.I. Artificial Intelligence – Wikipedia

Artificial Intelligence, China And The U.S. – How The U.S. Is …

National technology investment strategies are hard to define let alone pass through complicated legislative bodies, like the US Congress, even when theres a declared war that threatens a country’s financial and economic competitiveness.The war for global leadership in artificial intelligence and machine learning is well underway, and the US is poised to lose perhaps the most important technology war in its history.

Is the AI war well-understood?Not even close, at least not by the leaders who develop national strategies or by the citizens of the United States who all need to spend some time on https://willrobotstakemyjob.com.While I searched and searched, I could not find a single political candidate in the recent US mid-term elections who discussed AI, the AI war, or how the US will likely lose the war unless a massive strategic pivot occurs immediately.Since theyre mostly unaware of the war, US leaders have no strategies to prevent an historic loss: imagine the implications of electing politicians who have no idea a deadly war is underway.

The Threat

So whats going on?

AI/machine learning/deep learning (lets call it all AI) are the new digital weapons which, by the way, the US Department of Defense discovered decades ago.While we could certainly examine the importance of AI in global military and economic warfare, no one can argue that AI is unimportant.In fact, its at least a 9 or any imaginable 10-point scale.I give it an easy 10.So do lots of others who research technology trends and technology adoption, especially those who track indicators ofnational success.

The Chinese have a very public, very-deep, extremely well-funded commitment to AI.Air Force General VeraLinn Jamieson says it plainly:”We estimate the total spending on artificial intelligence systems in China in 2017 was $12 billion. We also estimate that it will grow to at least $70 billion by 2020.”According to the Obama White House Report in 2016, China publishes more journal articles on deep learning than the US and has increased its number of AI patents by 200%.China is determined to be the world leader in AI by 2030.

Listen to what Tristan Greene writing in TNW concludes about the USs commitment to AI:Unfortunately, despite congressional efforts to get the conversation started at the national level in the US, the White Houses current leadership doesnt appear interested in coming up with a strategy tokeep upwith China. It gets worse:China has allocated billions of dollars towards infrastructure to house hundreds of AI businesses in dedicated industrial parks.It has specific companies, the Chinese counterparts to US operations like Google and Amazon, working on different problems in the field of AI. And itsregulating education so that the nation produces more STEM workers. But perhaps most importantly, China makes it compulsory for businesses and private citizens to share their data with the government something far more valuable than money in the world of AI.

Greenes scary bottom line?Meanwhile, in the US, the Trump administration has shown little interest in discussing its own countrys AI yet,may soon have to talk to Chinas.

More data?According to Iris Deng, China ranks first in the quantity and citation of research papers, and holds the most AI patents, edging out the US and Japan (and) China has not been shy about its ambitions for AI dominance, with the State Council releasing a road map in July 2017 with a goal of creating a domestic industry worth 1 trillion yuan and becoming a global AI powerhouse by 2030.

It’s obvious:Without more leadership from Congress and the President, the U.S. is in serious danger of losing the economic and military rewards of artificial intelligence (AI) to China. Thats the somber conclusion of a report published … by the House Oversight and Reform IT subcommittee.

Jerry Bowles also says it clearly:The U.S. has traditionally led the world in the development and application of AI-driven technologies, due in part to the governments commitment to investing heavily in research and development. That has, in turn, helped support AIs growth and development. In 2015, the United States led the world in total gross domestic R&D expenditures, spending $497 billion.But, since then, neither Congress nor the Trump administration has paid much attention to AI and government R&D investment has been essentially flat.Meanwhile, China has made AI a key part of its formal economic plans for the future.

The Response

The US House of Representatives Subcommittee on Information Technology Committee on Oversight & Government Reform summarizes itbut notdefinitively:

There is a pressing need for conscious, direct, and spirited leadership from the Trump Administration.The 2016 reports put out by the Obama Administrations National Science and Technology Council and the recent actions of the Trump Administration are steps in the right direction. However, given the actions taken by other countries especially China Congress and the Administration will need to increase the time, attention, and level of resources the federal government devotes to AI research and development, as well as push for agencies to further build their capacities for adapting to advanced technologies.

The government has an essential role to play in securing American leadership in AI.Fulfilling this role will require balancing the creative energy of innovative Americans whose knowledge and entrepreneurial spirit have driven the development of this technology with regulatory frameworks that protect consumers. To ensure the appropriate balance is met, it is vital Congress and the Executive Branch continue to educate themselves about AI, increase the expenditures of R&D funds, help set the agenda for public debate, and, where appropriate, define the role of AI in the future of this nation.

Clearly, a coordinated, heavily-funded American response is way overdue.Here are somespecific steps:

These steps represent a good start to turn the tide of the AI war a war the US simply cannot afford to lose.

The rest is here:

Artificial Intelligence, China And The U.S. – How The U.S. Is …

A.I. Artificial Intelligence (2001) – IMDb

Edit Storyline

In the not-so-far future the polar ice caps have melted and the resulting rise of the ocean waters has drowned all the coastal cities of the world. Withdrawn to the interior of the continents, the human race keeps advancing, reaching the point of creating realistic robots (called mechas) to serve them. One of the mecha-producing companies builds David, an artificial kid which is the first to have real feelings, especially a never-ending love for his “mother”, Monica. Monica is the woman who adopted him as a substitute for her real son, who remains in cryo-stasis, stricken by an incurable disease. David is living happily with Monica and her husband, but when their real son returns home after a cure is discovered, his life changes dramatically. Written byChris Makrozahopoulos

Budget:$100,000,000 (estimated)

Opening Weekend USA: $29,352,630,1 July 2001, Wide Release

Gross USA: $78,616,689, 23 September 2001

Cumulative Worldwide Gross: $235,927,000

Runtime: 146 min

Aspect Ratio: 1.85 : 1

See the original post here:

A.I. Artificial Intelligence (2001) – IMDb

Artificial intelligence – Wikipedia

Intelligence demonstrated by machines

Artificial intelligence (AI), sometimes called machine intelligence, is intelligence demonstrated by machines, in contrast to the natural intelligence displayed by humans and other animals. In computer science AI research is defined as the study of “intelligent agents”: any device that perceives its environment and takes actions that maximize its chance of successfully achieving its goals.[1] Colloquially, the term “artificial intelligence” is applied when a machine mimics “cognitive” functions that humans associate with other human minds, such as “learning” and “problem solving”.

The scope of AI is disputed: as machines become increasingly capable, tasks considered as requiring “intelligence” are often removed from the definition, a phenomenon known as the AI effect, leading to the quip in Tesler’s Theorem, “AI is whatever hasn’t been done yet.”[3] For instance, optical character recognition is frequently excluded from “artificial intelligence”, having become a routine technology.[4] Modern machine capabilities generally classified as AI include successfully understanding human speech, competing at the highest level in strategic game systems (such as chess and Go),[6] autonomously operating cars, and intelligent routing in content delivery networks and military simulations.

Artificial intelligence was founded as an academic discipline in 1956, and in the years since has experienced several waves of optimism,[7][8] followed by disappointment and the loss of funding (known as an “AI winter”),[9][10] followed by new approaches, success and renewed funding.[8][11] For most of its history, AI research has been divided into subfields that often fail to communicate with each other.[12] These sub-fields are based on technical considerations, such as particular goals (e.g. “robotics” or “machine learning”),[13] the use of particular tools (“logic” or artificial neural networks), or deep philosophical differences.[14][15][16] Subfields have also been based on social factors (particular institutions or the work of particular researchers).[12]

The traditional problems (or goals) of AI research include reasoning, knowledge representation, planning, learning, natural language processing, perception and the ability to move and manipulate objects.[13] General intelligence is among the field’s long-term goals.[17] Approaches include statistical methods, computational intelligence, and traditional symbolic AI. Many tools are used in AI, including versions of search and mathematical optimization, artificial neural networks, and methods based on statistics, probability and economics. The AI field draws upon computer science, information engineering, mathematics, psychology, linguistics, philosophy, and many others.

The field was founded on the claim that human intelligence “can be so precisely described that a machine can be made to simulate it”.[18] This raises philosophical arguments about the nature of the mind and the ethics of creating artificial beings endowed with human-like intelligence which are issues that have been explored by myth, fiction and philosophy since antiquity.[19] Some people also consider AI to be a danger to humanity if it progresses unabated.[20] Others believe that AI, unlike previous technological revolutions, will create a risk of mass unemployment.[21]However Google is hosting a global contest to develop AI thats beneficial for humanity[22]

In the twenty-first century, AI techniques have experienced a resurgence following concurrent advances in computer power, large amounts of data, and theoretical understanding; and AI techniques have become an essential part of the technology industry, helping to solve many challenging problems in computer science, software engineering and operations research.[23][11]

Thought-capable artificial beings appeared as storytelling devices in antiquity,[24] and have been common in fiction, as in Mary Shelley’s Frankenstein or Karel apek’s R.U.R. (Rossum’s Universal Robots).[25] These characters and their fates raised many of the same issues now discussed in the ethics of artificial intelligence.[19]

The study of mechanical or “formal” reasoning began with philosophers and mathematicians in antiquity. The study of mathematical logic led directly to Alan Turing’s theory of computation, which suggested that a machine, by shuffling symbols as simple as “0” and “1”, could simulate any conceivable act of mathematical deduction. This insight, that digital computers can simulate any process of formal reasoning, is known as the ChurchTuring thesis.[26] Along with concurrent discoveries in neurobiology, information theory and cybernetics, this led researchers to consider the possibility of building an electronic brain. Turing proposed that “if a human could not distinguish between responses from a machine and a human, the machine could be considered intelligent”.[27] The first work that is now generally recognized as AI was McCullouch and Pitts’ 1943 formal design for Turing-complete “artificial neurons”.

The field of AI research was born at a workshop at Dartmouth College in 1956.[29] Attendees Allen Newell (CMU), Herbert Simon (CMU), John McCarthy (MIT), Marvin Minsky (MIT) and Arthur Samuel (IBM) became the founders and leaders of AI research.[30] They and their students produced programs that the press described as “astonishing”: computers were learning checkers strategies (c. 1954)[32] (and by 1959 were reportedly playing better than the average human),[33] solving word problems in algebra, proving logical theorems (Logic Theorist, first run c. 1956) and speaking English.[34] By the middle of the 1960s, research in the U.S. was heavily funded by the Department of Defense[35] and laboratories had been established around the world.[36] AI’s founders were optimistic about the future: Herbert Simon predicted, “machines will be capable, within twenty years, of doing any work a man can do”. Marvin Minsky agreed, writing, “within a generation… the problem of creating ‘artificial intelligence’ will substantially be solved”.[7]

They failed to recognize the difficulty of some of the remaining tasks. Progress slowed and in 1974, in response to the criticism of Sir James Lighthill and ongoing pressure from the US Congress to fund more productive projects, both the U.S. and British governments cut off exploratory research in AI. The next few years would later be called an “AI winter”,[9] a period when obtaining funding for AI projects was difficult.

In the early 1980s, AI research was revived by the commercial success of expert systems,[38] a form of AI program that simulated the knowledge and analytical skills of human experts. By 1985, the market for AI had reached over a billion dollars. At the same time, Japan’s fifth generation computer project inspired the U.S and British governments to restore funding for academic research.[8] However, beginning with the collapse of the Lisp Machine market in 1987, AI once again fell into disrepute, and a second, longer-lasting hiatus began.[10]

In the late 1990s and early 21st century, AI began to be used for logistics, data mining, medical diagnosis and other areas.[23] The success was due to increasing computational power (see Moore’s law), greater emphasis on solving specific problems, new ties between AI and other fields (such as statistics, economics and mathematics), and a commitment by researchers to mathematical methods and scientific standards.[39] Deep Blue became the first computer chess-playing system to beat a reigning world chess champion, Garry Kasparov, on 11 May 1997.

In 2011, a Jeopardy! quiz show exhibition match, IBM’s question answering system, Watson, defeated the two greatest Jeopardy! champions, Brad Rutter and Ken Jennings, by a significant margin. Faster computers, algorithmic improvements, and access to large amounts of data enabled advances in machine learning and perception; data-hungry deep learning methods started to dominate accuracy benchmarks around 2012.[42] The Kinect, which provides a 3D bodymotion interface for the Xbox 360 and the Xbox One, uses algorithms that emerged from lengthy AI research[43] as do intelligent personal assistants in smartphones.[44] In March 2016, AlphaGo won 4 out of 5 games of Go in a match with Go champion Lee Sedol, becoming the first computer Go-playing system to beat a professional Go player without handicaps.[6][45] In the 2017 Future of Go Summit, AlphaGo won a three-game match with Ke Jie,[46] who at the time continuously held the world No. 1 ranking for two years.[47][48] This marked the completion of a significant milestone in the development of Artificial Intelligence as Go is an extremely complex game, more so than Chess.

According to Bloomberg’s Jack Clark, 2015 was a landmark year for artificial intelligence, with the number of software projects that use AI within Google increased from a “sporadic usage” in 2012 to more than 2,700 projects. Clark also presents factual data indicating that error rates in image processing tasks have fallen significantly since 2011.[49] He attributes this to an increase in affordable neural networks, due to a rise in cloud computing infrastructure and to an increase in research tools and datasets.[11] Other cited examples include Microsoft’s development of a Skype system that can automatically translate from one language to another and Facebook’s system that can describe images to blind people.[49] In a 2017 survey, one in five companies reported they had “incorporated AI in some offerings or processes”.[50][51] Around 2016, China greatly accelerated its government funding; given its large supply of data and its rapidly increasing research output, some observers believe it may be on track to becoming an “AI superpower”.[52][53]

A typical AI perceives its environment and takes actions that maximize its chance of successfully achieving its goals.[1] An AI’s intended goal function can be simple (“1 if the AI wins a game of Go, 0 otherwise”) or complex (“Do actions mathematically similar to the actions that got you rewards in the past”). Goals can be explicitly defined, or can be induced. If the AI is programmed for “reinforcement learning”, goals can be implicitly induced by rewarding some types of behavior and punishing others.[a] Alternatively, an evolutionary system can induce goals by using a “fitness function” to mutate and preferentially replicate high-scoring AI systems; this is similar to how animals evolved to innately desire certain goals such as finding food, or how dogs can be bred via artificial selection to possess desired traits. Some AI systems, such as nearest-neighbor, instead reason by analogy; these systems are not generally given goals, except to the degree that goals are somehow implicit in their training data. Such systems can still be benchmarked if the non-goal system is framed as a system whose “goal” is to successfully accomplish its narrow classification task.[56]

AI often revolves around the use of algorithms. An algorithm is a set of unambiguous instructions that a mechanical computer can execute.[b] A complex algorithm is often built on top of other, simpler, algorithms. A simple example of an algorithm is the following recipe for optimal play at tic-tac-toe:

Many AI algorithms are capable of learning from data; they can enhance themselves by learning new heuristics (strategies, or “rules of thumb”, that have worked well in the past), or can themselves write other algorithms. Some of the “learners” described below, including Bayesian networks, decision trees, and nearest-neighbor, could theoretically, if given infinite data, time, and memory, learn to approximate any function, including whatever combination of mathematical functions would best describe the entire world. These learners could therefore, in theory, derive all possible knowledge, by considering every possible hypothesis and matching it against the data. In practice, it is almost never possible to consider every possibility, because of the phenomenon of “combinatorial explosion”, where the amount of time needed to solve a problem grows exponentially. Much of AI research involves figuring out how to identify and avoid considering broad swaths of possibilities that are unlikely to be fruitful.[58] For example, when viewing a map and looking for the shortest driving route from Denver to New York in the East, one can in most cases skip looking at any path through San Francisco or other areas far to the West; thus, an AI wielding an pathfinding algorithm like A* can avoid the combinatorial explosion that would ensue if every possible route had to be ponderously considered in turn.[60]

The earliest (and easiest to understand) approach to AI was symbolism (such as formal logic): “If an otherwise healthy adult has a fever, then they may have influenza”. A second, more general, approach is Bayesian inference: “If the current patient has a fever, adjust the probability they have influenza in such-and-such way”. The third major approach, extremely popular in routine business AI applications, are analogizers such as SVM and nearest-neighbor: “After examining the records of known past patients whose temperature, symptoms, age, and other factors mostly match the current patient, X% of those patients turned out to have influenza”. A fourth approach is harder to intuitively understand, but is inspired by how the brain’s machinery works: the artificial neural network approach uses artificial “neurons” that can learn by comparing itself to the desired output and altering the strengths of the connections between its internal neurons to “reinforce” connections that seemed to be useful. These four main approaches can overlap with each other and with evolutionary systems; for example, neural nets can learn to make inferences, to generalize, and to make analogies. Some systems implicitly or explicitly use multiple of these approaches, alongside many other AI and non-AI algorithms;[61] the best approach is often different depending on the problem.[63]

Learning algorithms work on the basis that strategies, algorithms, and inferences that worked well in the past are likely to continue working well in the future. These inferences can be obvious, such as “since the sun rose every morning for the last 10,000 days, it will probably rise tomorrow morning as well”. They can be nuanced, such as “X% of families have geographically separate species with color variants, so there is an Y% chance that undiscovered black swans exist”. Learners also work on the basis of “Occam’s razor”: The simplest theory that explains the data is the likeliest. Therefore, to be successful, a learner must be designed such that it prefers simpler theories to complex theories, except in cases where the complex theory is proven substantially better. Settling on a bad, overly complex theory gerrymandered to fit all the past training data is known as overfitting. Many systems attempt to reduce overfitting by rewarding a theory in accordance with how well it fits the data, but penalizing the theory in accordance with how complex the theory is. Besides classic overfitting, learners can also disappoint by “learning the wrong lesson”. A toy example is that an image classifier trained only on pictures of brown horses and black cats might conclude that all brown patches are likely to be horses. A real-world example is that, unlike humans, current image classifiers don’t determine the spatial relationship between components of the picture; instead, they learn abstract patterns of pixels that humans are oblivious to, but that linearly correlate with images of certain types of real objects. Faintly superimposing such a pattern on a legitimate image results in an “adversarial” image that the system misclassifies.[c][66][67][68]

Compared with humans, existing AI lacks several features of human “commonsense reasoning”; most notably, humans have powerful mechanisms for reasoning about “nave physics” such as space, time, and physical interactions. This enables even young children to easily make inferences like “If I roll this pen off a table, it will fall on the floor”. Humans also have a powerful mechanism of “folk psychology” that helps them to interpret natural-language sentences such as “The city councilmen refused the demonstrators a permit because they advocated violence”. (A generic AI has difficulty inferring whether the councilmen or the demonstrators are the ones alleged to be advocating violence.)[71][72][73] This lack of “common knowledge” means that AI often makes different mistakes than humans make, in ways that can seem incomprehensible. For example, existing self-driving cars cannot reason about the location nor the intentions of pedestrians in the exact way that humans do, and instead must use non-human modes of reasoning to avoid accidents.[74][75][76]

The overall research goal of artificial intelligence is to create technology that allows computers and machines to function in an intelligent manner. The general problem of simulating (or creating) intelligence has been broken down into sub-problems. These consist of particular traits or capabilities that researchers expect an intelligent system to display. The traits described below have received the most attention.[13]

Early researchers developed algorithms that imitated step-by-step reasoning that humans use when they solve puzzles or make logical deductions.[77] By the late 1980s and 1990s, AI research had developed methods for dealing with uncertain or incomplete information, employing concepts from probability and economics.[78]

These algorithms proved to be insufficient for solving large reasoning problems, because they experienced a “combinatorial explosion”: they became exponentially slower as the problems grew larger.[58] In fact, even humans rarely use the step-by-step deduction that early AI research was able to model. They solve most of their problems using fast, intuitive judgements.[79]

Knowledge representation[80] and knowledge engineering[81] are central to classical AI research. Some “expert systems” attempt to gather together explicit knowledge possessed by experts in some narrow domain. In addition, some projects attempt to gather the “commonsense knowledge” known to the average person into a database containing extensive knowledge about the world. Among the things a comprehensive commonsense knowledge base would contain are: objects, properties, categories and relations between objects;[82] situations, events, states and time;[83] causes and effects;[84] knowledge about knowledge (what we know about what other people know);[85] and many other, less well researched domains. A representation of “what exists” is an ontology: the set of objects, relations, concepts, and properties formally described so that software agents can interpret them. The semantics of these are captured as description logic concepts, roles, and individuals, and typically implemented as classes, properties, and individuals in the Web Ontology Language.[86] The most general ontologies are called upper ontologies, which attempt to provide a foundation for all other knowledge[87] by acting as mediators between domain ontologies that cover specific knowledge about a particular knowledge domain (field of interest or area of concern). Such formal knowledge representations can be used in content-based indexing and retrieval,[88] scene interpretation,[89] clinical decision support,[90] knowledge discovery (mining “interesting” and actionable inferences from large databases),[91] and other areas.[92]

Among the most difficult problems in knowledge representation are:

Intelligent agents must be able to set goals and achieve them.[99] They need a way to visualize the futurea representation of the state of the world and be able to make predictions about how their actions will change itand be able to make choices that maximize the utility (or “value”) of available choices.[100]

In classical planning problems, the agent can assume that it is the only system acting in the world, allowing the agent to be certain of the consequences of its actions.[101] However, if the agent is not the only actor, then it requires that the agent can reason under uncertainty. This calls for an agent that can not only assess its environment and make predictions, but also evaluate its predictions and adapt based on its assessment.[102]

Multi-agent planning uses the cooperation and competition of many agents to achieve a given goal. Emergent behavior such as this is used by evolutionary algorithms and swarm intelligence.[103]

Machine learning, a fundamental concept of AI research since the field’s inception,[104] is the study of computer algorithms that improve automatically through experience.[105][106]

Unsupervised learning is the ability to find patterns in a stream of input, without requiring a human to label the inputs first.[107] Supervised learning includes both classification and numerical regression, which requires a human to label the input data first. Classification is used to determine what category something belongs in, after seeing a number of examples of things from several categories. Regression is the attempt to produce a function that describes the relationship between inputs and outputs and predicts how the outputs should change as the inputs change.[106] Both classifiers and regression learners can be viewed as “function approximators” trying to learn an unknown (possibly implicit) function; for example, a spam classifier can be viewed as learning a function that maps from the text of an email to one of two categories, “spam” or “not spam”. Computational learning theory can assess learners by computational complexity, by sample complexity (how much data is required), or by other notions of optimization.[108] In reinforcement learning[109] the agent is rewarded for good responses and punished for bad ones. The agent uses this sequence of rewards and punishments to form a strategy for operating in its problem space.

Natural language processing[110] (NLP) gives machines the ability to read and understand human language. A sufficiently powerful natural language processing system would enable natural-language user interfaces and the acquisition of knowledge directly from human-written sources, such as newswire texts. Some straightforward applications of natural language processing include information retrieval, text mining, question answering[111] and machine translation.[112] Many current approaches use word co-occurrence frequencies to construct syntactic representations of text. “Keyword spotting” strategies for search are popular and scalable but dumb; a search query for “dog” might only match documents with the literal word “dog” and miss a document with the word “poodle”. “Lexical affinity” strategies use the occurrence of words such as “accident” to assess the sentiment of a document. Modern statistical NLP approaches can combine all these strategies as well as others, and often achieve acceptable accuracy at the page or paragraph level, but continue to lack the semantic understanding required to classify isolated sentences well. Besides the usual difficulties with encoding semantic commonsense knowledge, existing semantic NLP sometimes scales too poorly to be viable in business applications. Beyond semantic NLP, the ultimate goal of “narrative” NLP is to embody a full understanding of commonsense reasoning.[113]

Machine perception[114] is the ability to use input from sensors (such as cameras (visible spectrum or infrared), microphones, wireless signals, and active lidar, sonar, radar, and tactile sensors) to deduce aspects of the world. Applications include speech recognition,[115] facial recognition, and object recognition.[116] Computer vision is the ability to analyze visual input. Such input is usually ambiguous; a giant, fifty-meter-tall pedestrian far away may produce exactly the same pixels as a nearby normal-sized pedestrian, requiring the AI to judge the relative likelihood and reasonableness of different interpretations, for example by using its “object model” to assess that fifty-meter pedestrians do not exist.[117]

AI is heavily used in robotics.[118] Advanced robotic arms and other industrial robots, widely used in modern factories, can learn from experience how to move efficiently despite the presence of friction and gear slippage.[119] A modern mobile robot, when given a small, static, and visible environment, can easily determine its location and map its environment; however, dynamic environments, such as (in endoscopy) the interior of a patient’s breathing body, pose a greater challenge. Motion planning is the process of breaking down a movement task into “primitives” such as individual joint movements. Such movement often involves compliant motion, a process where movement requires maintaining physical contact with an object.[121][122] Moravec’s paradox generalizes that low-level sensorimotor skills that humans take for granted are, counterintuitively, difficult to program into a robot; the paradox is named after Hans Moravec, who stated in 1988 that “it is comparatively easy to make computers exhibit adult level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility”.[123][124] This is attributed to the fact that, unlike checkers, physical dexterity has been a direct target of natural selection for millions of years.[125]

Moravec’s paradox can be extended to many forms of social intelligence.[127][128] Distributed multi-agent coordination of autonomous vehicles remains a difficult problem.[129] Affective computing is an interdisciplinary umbrella that comprises systems which recognize, interpret, process, or simulate human affects. Moderate successes related to affective computing include textual sentiment analysis and, more recently, multimodal affect analysis (see multimodal sentiment analysis), wherein AI classifies the affects displayed by a videotaped subject.[133]

In the long run, social skills and an understanding of human emotion and game theory would be valuable to a social agent. Being able to predict the actions of others by understanding their motives and emotional states would allow an agent to make better decisions. Some computer systems mimic human emotion and expressions to appear more sensitive to the emotional dynamics of human interaction, or to otherwise facilitate humancomputer interaction.[134] Similarly, some virtual assistants are programmed to speak conversationally or even to banter humorously; this tends to give nave users an unrealistic conception of how intelligent existing computer agents actually are.[135]

Historically, projects such as the Cyc knowledge base (1984) and the massive Japanese Fifth Generation Computer Systems initiative (19821992) attempted to cover the breadth of human cognition. These early projects failed to escape the limitations of non-quantitative symbolic logic models and, in retrospect, greatly underestimated the difficulty of cross-domain AI. Nowadays, the vast majority of current AI researchers work instead on tractable “narrow AI” applications (such as medical diagnosis or automobile navigation).[136] Many researchers predict that such “narrow AI” work in different individual domains will eventually be incorporated into a machine with artificial general intelligence (AGI), combining most of the narrow skills mentioned in this article and at some point even exceeding human ability in most or all these areas.[17][137] Many advances have general, cross-domain significance. One high-profile example is that DeepMind in the 2010s developed a “generalized artificial intelligence” that could learn many diverse Atari games on its own, and later developed a variant of the system which succeeds at sequential learning.[138][139][140] Besides transfer learning,[141] hypothetical AGI breakthroughs could include the development of reflective architectures that can engage in decision-theoretic metareasoning, and figuring out how to “slurp up” a comprehensive knowledge base from the entire unstructured Web. Some argue that some kind of (currently-undiscovered) conceptually straightforward, but mathematically difficult, “Master Algorithm” could lead to AGI. Finally, a few “emergent” approaches look to simulating human intelligence extremely closely, and believe that anthropomorphic features like an artificial brain or simulated child development may someday reach a critical point where general intelligence emerges.[143][144]

Many of the problems in this article may also require general intelligence, if machines are to solve the problems as well as people do. For example, even specific straightforward tasks, like machine translation, require that a machine read and write in both languages (NLP), follow the author’s argument (reason), know what is being talked about (knowledge), and faithfully reproduce the author’s original intent (social intelligence). A problem like machine translation is considered “AI-complete”, because all of these problems need to be solved simultaneously in order to reach human-level machine performance.

There is no established unifying theory or paradigm that guides AI research. Researchers disagree about many issues.[145] A few of the most long standing questions that have remained unanswered are these: should artificial intelligence simulate natural intelligence by studying psychology or neurobiology? Or is human biology as irrelevant to AI research as bird biology is to aeronautical engineering?[14]Can intelligent behavior be described using simple, elegant principles (such as logic or optimization)? Or does it necessarily require solving a large number of completely unrelated problems?[15]

In the 1940s and 1950s, a number of researchers explored the connection between neurobiology, information theory, and cybernetics. Some of them built machines that used electronic networks to exhibit rudimentary intelligence, such as W. Grey Walter’s turtles and the Johns Hopkins Beast. Many of these researchers gathered for meetings of the Teleological Society at Princeton University and the Ratio Club in England.[146] By 1960, this approach was largely abandoned, although elements of it would be revived in the 1980s.

When access to digital computers became possible in the middle 1950s, AI research began to explore the possibility that human intelligence could be reduced to symbol manipulation. The research was centered in three institutions: Carnegie Mellon University, Stanford and MIT, and as described below, each one developed its own style of research. John Haugeland named these symbolic approaches to AI “good old fashioned AI” or “GOFAI”.[147] During the 1960s, symbolic approaches had achieved great success at simulating high-level thinking in small demonstration programs. Approaches based on cybernetics or artificial neural networks were abandoned or pushed into the background.[148]Researchers in the 1960s and the 1970s were convinced that symbolic approaches would eventually succeed in creating a machine with artificial general intelligence and considered this the goal of their field.

Economist Herbert Simon and Allen Newell studied human problem-solving skills and attempted to formalize them, and their work laid the foundations of the field of artificial intelligence, as well as cognitive science, operations research and management science. Their research team used the results of psychological experiments to develop programs that simulated the techniques that people used to solve problems. This tradition, centered at Carnegie Mellon University would eventually culminate in the development of the Soar architecture in the middle 1980s.[149][150]

Unlike Simon and Newell, John McCarthy felt that machines did not need to simulate human thought, but should instead try to find the essence of abstract reasoning and problem-solving, regardless of whether people used the same algorithms.[14] His laboratory at Stanford (SAIL) focused on using formal logic to solve a wide variety of problems, including knowledge representation, planning and learning.[151] Logic was also the focus of the work at the University of Edinburgh and elsewhere in Europe which led to the development of the programming language Prolog and the science of logic programming.[152]

Researchers at MIT (such as Marvin Minsky and Seymour Papert)[153] found that solving difficult problems in vision and natural language processing required ad-hoc solutions they argued that there was no simple and general principle (like logic) that would capture all the aspects of intelligent behavior. Roger Schank described their “anti-logic” approaches as “scruffy” (as opposed to the “neat” paradigms at CMU and Stanford).[15] Commonsense knowledge bases (such as Doug Lenat’s Cyc) are an example of “scruffy” AI, since they must be built by hand, one complicated concept at a time.[154]

When computers with large memories became available around 1970, researchers from all three traditions began to build knowledge into AI applications.[155] This “knowledge revolution” led to the development and deployment of expert systems (introduced by Edward Feigenbaum), the first truly successful form of AI software.[38] Key component on system arhitecute for all expert systems is Knowledge base, which stores facts and rules that illustrates AI.[156] The knowledge revolution was also driven by the realization that enormous amounts of knowledge would be required by many simple AI applications.

By the 1980s, progress in symbolic AI seemed to stall and many believed that symbolic systems would never be able to imitate all the processes of human cognition, especially perception, robotics, learning and pattern recognition. A number of researchers began to look into “sub-symbolic” approaches to specific AI problems.[16] Sub-symbolic methods manage to approach intelligence without specific representations of knowledge.

This includes embodied, situated, behavior-based, and nouvelle AI. Researchers from the related field of robotics, such as Rodney Brooks, rejected symbolic AI and focused on the basic engineering problems that would allow robots to move and survive.[157] Their work revived the non-symbolic viewpoint of the early cybernetics researchers of the 1950s and reintroduced the use of control theory in AI. This coincided with the development of the embodied mind thesis in the related field of cognitive science: the idea that aspects of the body (such as movement, perception and visualization) are required for higher intelligence.

Within developmental robotics, developmental learning approaches are elaborated upon to allow robots to accumulate repertoires of novel skills through autonomous self-exploration, social interaction with human teachers, and the use of guidance mechanisms (active learning, maturation, motor synergies, etc.).[158][159]

Interest in neural networks and “connectionism” was revived by David Rumelhart and others in the middle of the 1980s.[162] Artificial neural networks are an example of soft computing — they are solutions to problems which cannot be solved with complete logical certainty, and where an approximate solution is often sufficient. Other soft computing approaches to AI include fuzzy systems, evolutionary computation and many statistical tools. The application of soft computing to AI is studied collectively by the emerging discipline of computational intelligence.[163]

Much of traditional GOFAI got bogged down on ad hoc patches to symbolic computation that worked on their own toy models but failed to generalize to real-world results. However, around the 1990s, AI researchers adopted sophisticated mathematical tools, such as hidden Markov models (HMM), information theory, and normative Bayesian decision theory to compare or to unify competing architectures. The shared mathematical language permitted a high level of collaboration with more established fields (like mathematics, economics or operations research).[d] Compared with GOFAI, new “statistical learning” techniques such as HMM and neural networks were gaining higher levels of accuracy in many practical domains such as data mining, without necessarily acquiring semantic understanding of the datasets. The increased successes with real-world data led to increasing emphasis on comparing different approaches against shared test data to see which approach performed best in a broader context than that provided by idiosyncratic toy models; AI research was becoming more scientific. Nowadays results of experiments are often rigorously measurable, and are sometimes (with difficulty) reproducible.[39][164] Different statistical learning techniques have different limitations; for example, basic HMM cannot model the infinite possible combinations of natural language. Critics note that the shift from GOFAI to statistical learning is often also a shift away from Explainable AI. In AGI research, some scholars caution against over-reliance on statistical learning, and argue that continuing research into GOFAI will still be necessary to attain general intelligence.

AI has developed a large number of tools to solve the most difficult problems in computer science. A few of the most general of these methods are discussed below.

Many problems in AI can be solved in theory by intelligently searching through many possible solutions:[173] Reasoning can be reduced to performing a search. For example, logical proof can be viewed as searching for a path that leads from premises to conclusions, where each step is the application of an inference rule.[174] Planning algorithms search through trees of goals and subgoals, attempting to find a path to a target goal, a process called means-ends analysis.[175] Robotics algorithms for moving limbs and grasping objects use local searches in configuration space.[119] Many learning algorithms use search algorithms based on optimization.

Simple exhaustive searches[176] are rarely sufficient for most real-world problems: the search space (the number of places to search) quickly grows to astronomical numbers. The result is a search that is too slow or never completes. The solution, for many problems, is to use “heuristics” or “rules of thumb” that prioritize choices in favor of those that are more likely to reach a goal and to do so in a shorter number of steps. In some search methodologies heuristics can also serve to entirely eliminate some choices that are unlikely to lead to a goal (called “pruning the search tree”). Heuristics supply the program with a “best guess” for the path on which the solution lies.[177] Heuristics limit the search for solutions into a smaller sample size.

A very different kind of search came to prominence in the 1990s, based on the mathematical theory of optimization. For many problems, it is possible to begin the search with some form of a guess and then refine the guess incrementally until no more refinements can be made. These algorithms can be visualized as blind hill climbing: we begin the search at a random point on the landscape, and then, by jumps or steps, we keep moving our guess uphill, until we reach the top. Other optimization algorithms are simulated annealing, beam search and random optimization.[178]

Evolutionary computation uses a form of optimization search. For example, they may begin with a population of organisms (the guesses) and then allow them to mutate and recombine, selecting only the fittest to survive each generation (refining the guesses). Classic evolutionary algorithms include genetic algorithms, gene expression programming, and genetic programming.[179] Alternatively, distributed search processes can coordinate via swarm intelligence algorithms. Two popular swarm algorithms used in search are particle swarm optimization (inspired by bird flocking) and ant colony optimization (inspired by ant trails).[180][181]

Logic[182] is used for knowledge representation and problem solving, but it can be applied to other problems as well. For example, the satplan algorithm uses logic for planning[183] and inductive logic programming is a method for learning.[184]

Several different forms of logic are used in AI research. Propositional logic[185] involves truth functions such as “or” and “not”. First-order logic[186] adds quantifiers and predicates, and can express facts about objects, their properties, and their relations with each other. Fuzzy set theory assigns a “degree of truth” (between 0 and 1) to vague statements such as “Alice is old” (or rich, or tall, or hungry) that are too linguistically imprecise to be completely true or false. Fuzzy logic is successfully used in control systems to allow experts to contribute vague rules such as “if you are close to the destination station and moving fast, increase the train’s brake pressure”; these vague rules can then be numerically refined within the system. Fuzzy logic fails to scale well in knowledge bases; many AI researchers question the validity of chaining fuzzy-logic inferences.[e][188][189]

Default logics, non-monotonic logics and circumscription[94] are forms of logic designed to help with default reasoning and the qualification problem. Several extensions of logic have been designed to handle specific domains of knowledge, such as: description logics;[82] situation calculus, event calculus and fluent calculus (for representing events and time);[83] causal calculus;[84] belief calculus;[190] and modal logics.[85]

Overall, qualitiative symbolic logic is brittle and scales poorly in the presence of noise or other uncertainty. Exceptions to rules are numerous, and it is difficult for logical systems to function in the presence of contradictory rules.[192]

Many problems in AI (in reasoning, planning, learning, perception, and robotics) require the agent to operate with incomplete or uncertain information. AI researchers have devised a number of powerful tools to solve these problems using methods from probability theory and economics.[193]

Bayesian networks[194] are a very general tool that can be used for a large number of problems: reasoning (using the Bayesian inference algorithm),[195] learning (using the expectation-maximization algorithm),[f][197] planning (using decision networks)[198] and perception (using dynamic Bayesian networks).[199] Probabilistic algorithms can also be used for filtering, prediction, smoothing and finding explanations for streams of data, helping perception systems to analyze processes that occur over time (e.g., hidden Markov models or Kalman filters).[199] Compared with symbolic logic, formal Bayesian inference is computationally expensive. For inference to be tractable, most observations must be conditionally independent of one another. Complicated graphs with diamonds or other “loops” (undirected cycles) can require a sophisticated method such as Markov Chain Monte Carlo, which spreads an ensemble of random walkers throughout the Bayesian network and attempts to converge to an assessment of the conditional probabilities. Bayesian networks are used on Xbox Live to rate and match players; wins and losses are “evidence” of how good a player is. AdSense uses a Bayesian network with over 300 million edges to learn which ads to serve.

A key concept from the science of economics is “utility”: a measure of how valuable something is to an intelligent agent. Precise mathematical tools have been developed that analyze how an agent can make choices and plan, using decision theory, decision analysis,[200] and information value theory.[100] These tools include models such as Markov decision processes,[201] dynamic decision networks,[199] game theory and mechanism design.[202]

The simplest AI applications can be divided into two types: classifiers (“if shiny then diamond”) and controllers (“if shiny then pick up”). Controllers do, however, also classify conditions before inferring actions, and therefore classification forms a central part of many AI systems. Classifiers are functions that use pattern matching to determine a closest match. They can be tuned according to examples, making them very attractive for use in AI. These examples are known as observations or patterns. In supervised learning, each pattern belongs to a certain predefined class. A class can be seen as a decision that has to be made. All the observations combined with their class labels are known as a data set. When a new observation is received, that observation is classified based on previous experience.[203]

A classifier can be trained in various ways; there are many statistical and machine learning approaches. The decision tree[204] is perhaps the most widely used machine learning algorithm. Other widely used classifiers are the neural network,[206]k-nearest neighbor algorithm,[g][208]kernel methods such as the support vector machine (SVM),[h][210]Gaussian mixture model,[211] and the extremely popular naive Bayes classifier.[i][213] Classifier performance depends greatly on the characteristics of the data to be classified, such as the dataset size, the dimensionality, and the level of noise. Model-based classifiers perform well if the assumed model is an extremely good fit for the actual data. Otherwise, if no matching model is available, and if accuracy (rather than speed or scalability) is the sole concern, conventional wisdom is that discriminative classifiers (especially SVM) tend to be more accurate than model-based classifiers such as “naive Bayes” on most practical data sets.[214]

Neural networks, or neural nets, were inspired by the architecture of neurons in the human brain. A simple “neuron” N accepts input from multiple other neurons, each of which, when activated (or “fired”), cast a weighted “vote” for or against whether neuron N should itself activate. Learning requires an algorithm to adjust these weights based on the training data; one simple algorithm (dubbed “fire together, wire together”) is to increase the weight between two connected neurons when the activation of one triggers the successful activation of another. The net forms “concepts” that are distributed among a subnetwork of shared[j] neurons that tend to fire together; a concept meaning “leg” might be coupled with a subnetwork meaning “foot” that includes the sound for “foot”. Neurons have a continuous spectrum of activation; in addition, neurons can process inputs in a nonlinear way rather than weighing straightforward votes. Modern neural nets can learn both continuous functions and, surprisingly, digital logical operations. Neural networks’ early successes included predicting the stock market and (in 1995) a mostly self-driving car.[k] In the 2010s, advances in neural networks using deep learning thrust AI into widespread public consciousness and contributed to an enormous upshift in corporate AI spending; for example, AI-related M&A in 2017 was over 25 times as large as in 2015.[217][218]

The study of non-learning artificial neural networks[206] began in the decade before the field of AI research was founded, in the work of Walter Pitts and Warren McCullouch. Frank Rosenblatt invented the perceptron, a learning network with a single layer, similar to the old concept of linear regression. Early pioneers also include Alexey Grigorevich Ivakhnenko, Teuvo Kohonen, Stephen Grossberg, Kunihiko Fukushima, Christoph von der Malsburg, David Willshaw, Shun-Ichi Amari, Bernard Widrow, John Hopfield, Eduardo R. Caianiello, and others.

The main categories of networks are acyclic or feedforward neural networks (where the signal passes in only one direction) and recurrent neural networks (which allow feedback and short-term memories of previous input events). Among the most popular feedforward networks are perceptrons, multi-layer perceptrons and radial basis networks.[219] Neural networks can be applied to the problem of intelligent control (for robotics) or learning, using such techniques as Hebbian learning (“fire together, wire together”), GMDH or competitive learning.[220]

Today, neural networks are often trained by the backpropagation algorithm, which had been around since 1970 as the reverse mode of automatic differentiation published by Seppo Linnainmaa,[221][222] and was introduced to neural networks by Paul Werbos.[223][224][225]

Hierarchical temporal memory is an approach that models some of the structural and algorithmic properties of the neocortex.[226]

In short, most neural networks use some form of gradient descent on a hand-created neural topology. However, some research groups, such as Uber, argue that simple neuroevolution to mutate new neural network topologies and weights may be competitive with sophisticated gradient descent approaches. One advantage of neuroevolution is that it may be less prone to get caught in “dead ends”.[227]

Deep learning is any artificial neural network that can learn a long chain of causal links. For example, a feedforward network with six hidden layers can learn a seven-link causal chain (six hidden layers + output layer) and has a “credit assignment path” (CAP) depth of seven. Many deep learning systems need to be able to learn chains ten or more causal links in length.[228] Deep learning has transformed many important subfields of artificial intelligence, including computer vision, speech recognition, natural language processing and others.[229][230][228]

According to one overview,[231] the expression “Deep Learning” was introduced to the Machine Learning community by Rina Dechter in 1986[232] and gained traction afterIgor Aizenberg and colleagues introduced it to Artificial Neural Networks in 2000.[233] The first functional Deep Learning networks were published by Alexey Grigorevich Ivakhnenko and V. G. Lapa in 1965.[234][pageneeded] These networks are trained one layer at a time. Ivakhnenko’s 1971 paper[235] describes the learning of a deep feedforward multilayer perceptron with eight layers, already much deeper than many later networks. In 2006, a publication by Geoffrey Hinton and Ruslan Salakhutdinov introduced another way of pre-training many-layered feedforward neural networks (FNNs) one layer at a time, treating each layer in turn as an unsupervised restricted Boltzmann machine, then using supervised backpropagation for fine-tuning. Similar to shallow artificial neural networks, deep neural networks can model complex non-linear relationships. Over the last few years, advances in both machine learning algorithms and computer hardware have led to more efficient methods for training deep neural networks that contain many layers of non-linear hidden units and a very large output layer.[237]

Deep learning often uses convolutional neural networks (CNNs), whose origins can be traced back to the Neocognitron introduced by Kunihiko Fukushima in 1980.[238] In 1989, Yann LeCun and colleagues applied backpropagation to such an architecture. In the early 2000s, in an industrial application CNNs already processed an estimated 10% to 20% of all the checks written in the US.[239]Since 2011, fast implementations of CNNs on GPUs havewon many visual pattern recognition competitions.[228]

CNNs with 12 convolutional layers were used in conjunction with reinforcement learning by Deepmind’s “AlphaGo Lee”, the program that beat a top Go champion in 2016.[240]

Early on, deep learning was also applied to sequence learning with recurrent neural networks (RNNs)[241] which are in theory Turing complete[242] and can run arbitrary programs to process arbitrary sequences of inputs. The depth of an RNN is unlimited and depends on the length of its input sequence; thus, an RNN is an example of deep learning.[228] RNNs can be trained by gradient descent[243][244][245] but suffer from the vanishing gradient problem.[229][246] In 1992, it was shown that unsupervised pre-training of a stack of recurrent neural networks can speed up subsequent supervised learning of deep sequential problems.[247]

Numerous researchers now use variants of a deep learning recurrent NN called the long short-term memory (LSTM) network published by Hochreiter & Schmidhuber in 1997.[248] LSTM is often trained by Connectionist Temporal Classification (CTC).[249] At Google, Microsoft and Baidu this approach has revolutionised speech recognition.[250][251][252] For example, in 2015, Google’s speech recognition experienced a dramatic performance jump of 49% through CTC-trained LSTM, which is now available through Google Voice to billions of smartphone users.[253] Google also used LSTM to improve machine translation,[254] Language Modeling[255] and Multilingual Language Processing.[256] LSTM combined with CNNs also improved automatic image captioning[257] and a plethora of other applications.

AI, like electricity or the steam engine, is a general purpose technology. There is no consensus on how to characterize which tasks AI tends to excel at.[258] While projects such as AlphaZero have succeeded in generating their own knowledge from scratch, many other machine learning projects require large training datasets.[259][260] Researcher Andrew Ng has suggested, as a “highly imperfect rule of thumb”, that “almost anything a typical human can do with less than one second of mental thought, we can probably now or in the near future automate using AI.”[261] Moravec’s paradox suggests that AI lags humans at many tasks that the human brain has specifically evolved to perform well.[125]

Games provide a well-publicized benchmark for assessing rates of progress. AlphaGo around 2016 brought the era of classical board-game benchmarks to a close. Games of imperfect knowledge provide new challenges to AI in the area of game theory.[262][263] E-sports such as StarCraft continue to provide additional public benchmarks.[264][265] There are many competitions and prizes, such as the Imagenet Challenge, to promote research in artificial intelligence. The most common areas of competition include general machine intelligence, conversational behavior, data-mining, robotic cars, and robot soccer as well as conventional games.[266]

The “imitation game” (an interpretation of the 1950 Turing test that assesses whether a computer can imitate a human) is nowadays considered too exploitable to be a meaningful benchmark.[267] A derivative of the Turing test is the Completely Automated Public Turing test to tell Computers and Humans Apart (CAPTCHA). As the name implies, this helps to determine that a user is an actual person and not a computer posing as a human. In contrast to the standard Turing test, CAPTCHA is administered by a machine and targeted to a human as opposed to being administered by a human and targeted to a machine. A computer asks a user to complete a simple test then generates a grade for that test. Computers are unable to solve the problem, so correct solutions are deemed to be the result of a person taking the test. A common type of CAPTCHA is the test that requires the typing of distorted letters, numbers or symbols that appear in an image undecipherable by a computer.

Proposed “universal intelligence” tests aim to compare how well machines, humans, and even non-human animals perform on problem sets that are generic as possible. At an extreme, the test suite can contain every possible problem, weighted by Kolmogorov complexity; unfortunately, these problem sets tend to be dominated by impoverished pattern-matching exercises where a tuned AI can easily exceed human performance levels.[269][270]

AI is relevant to any intellectual task. Modern artificial intelligence techniques are pervasive and are too numerous to list here. Frequently, when a technique reaches mainstream use, it is no longer considered artificial intelligence; this phenomenon is described as the AI effect.

High-profile examples of AI include autonomous vehicles (such as drones and self-driving cars), medical diagnosis, creating art (such as poetry), proving mathematical theorems, playing games (such as Chess or Go), search engines (such as Google search), online assistants (such as Siri), image recognition in photographs, spam filtering, prediction of judicial decisions[273] and targeting online advertisements.[274][275]

With social media sites overtaking TV as a source for news for young people and news organisations increasingly reliant on social media platforms for generating distribution,[276] major publishers now use artificial intelligence (AI) technology to post stories more effectively and generate higher volumes of traffic.[277]

AI is being applied to the high cost problem of dosage issueswhere findings suggested that AI could save $16 billion. In 2016, a ground breaking study in California found that a mathematical formula developed with the help of AI correctly determined the accurate dose of immunosuppressant drugs to give to organ patients.[278]

Artificial intelligence is breaking into the healthcare industry by assisting doctors. According to Bloomberg Technology, Microsoft has developed AI to help doctors find the right treatments for cancer.[279] There is a great amount of research and drugs developed relating to cancer. In detail, there are more than 800 medicines and vaccines to treat cancer. This negatively affects the doctors, because there are too many options to choose from, making it more difficult to choose the right drugs for the patients. Microsoft is working on a project to develop a machine called “Hanover”. Its goal is to memorize all the papers necessary to cancer and help predict which combinations of drugs will be most effective for each patient. One project that is being worked on at the moment is fighting myeloid leukemia, a fatal cancer where the treatment has not improved in decades. Another study was reported to have found that artificial intelligence was as good as trained doctors in identifying skin cancers.[280] Another study is using artificial intelligence to try and monitor multiple high-risk patients, and this is done by asking each patient numerous questions based on data acquired from live doctor to patient interactions.[281]

According to CNN, a recent study by surgeons at the Children’s National Medical Center in Washington successfully demonstrated surgery with an autonomous robot. The team supervised the robot while it performed soft-tissue surgery, stitching together a pig’s bowel during open surgery, and doing so better than a human surgeon, the team claimed.[282] IBM has created its own artificial intelligence computer, the IBM Watson, which has beaten human intelligence (at some levels). Watson not only won at the game show Jeopardy! against former champions,[283] but was declared a hero after successfully diagnosing a woman who was suffering from leukemia.[284]

Advancements in AI have contributed to the growth of the automotive industry through the creation and evolution of self-driving vehicles. As of 2016[update], there are over 30 companies utilizing AI into the creation of driverless cars. A few companies involved with AI include Tesla, Google, and Apple.[285]

Many components contribute to the functioning of self-driving cars. These vehicles incorporate systems such as braking, lane changing, collision prevention, navigation and mapping. Together, these systems, as well as high performance computers, are integrated into one complex vehicle.[286]

Recent developments in autonomous automobiles have made the innovation of self-driving trucks possible, though they are still in the testing phase. The UK government has passed legislation to begin testing of self-driving truck platoons in 2018.[287] Self-driving truck platoons are a fleet of self-driving trucks following the lead of one non-self-driving truck, so the truck platoons aren’t entirely autonomous yet. Meanwhile, the Daimler, a German automobile corporation, is testing the Freightliner Inspiration which is a semi-autonomous truck that will only be used on the highway.[288]

One main factor that influences the ability for a driver-less automobile to function is mapping. In general, the vehicle would be pre-programmed with a map of the area being driven. This map would include data on the approximations of street light and curb heights in order for the vehicle to be aware of its surroundings. However, Google has been working on an algorithm with the purpose of eliminating the need for pre-programmed maps and instead, creating a device that would be able to adjust to a variety of new surroundings.[289] Some self-driving cars are not equipped with steering wheels or brake pedals, so there has also been research focused on creating an algorithm that is capable of maintaining a safe environment for the passengers in the vehicle through awareness of speed and driving conditions.[290]

Another factor that is influencing the ability for a driver-less automobile is the safety of the passenger. To make a driver-less automobile, engineers must program it to handle high-risk situations. These situations could include a head-on collision with pedestrians. The car’s main goal should be to make a decision that would avoid hitting the pedestrians and saving the passengers in the car. But there is a possibility the car would need to make a decision that would put someone in danger. In other words, the car would need to decide to save the pedestrians or the passengers.[291] The programming of the car in these situations is crucial to a successful driver-less automobile.

Financial institutions have long used artificial neural network systems to detect charges or claims outside of the norm, flagging these for human investigation. The use of AI in banking can be traced back to 1987 when Security Pacific National Bank in US set-up a Fraud Prevention Task force to counter the unauthorised use of debit cards. Programs like Kasisto and Moneystream are using AI in financial services.

Banks use artificial intelligence systems today to organize operations, maintain book-keeping, invest in stocks, and manage properties. AI can react to changes overnight or when business is not taking place.[292] In August 2001, robots beat humans in a simulated financial trading competition.[293] AI has also reduced fraud and financial crimes by monitoring behavioral patterns of users for any abnormal changes or anomalies.[294]

The use of AI machines in the market in applications such as online trading and decision making has changed major economic theories.[295] For example, AI based buying and selling platforms have changed the law of supply and demand in that it is now possible to easily estimate individualized demand and supply curves and thus individualized pricing. Furthermore, AI machines reduce information asymmetry in the market and thus making markets more efficient while reducing the volume of trades. Furthermore, AI in the markets limits the consequences of behavior in the markets again making markets more efficient. Other theories where AI has had impact include in rational choice, rational expectations, game theory, Lewis turning point, portfolio optimization and counterfactual thinking.

In video games, artificial intelligence is routinely used to generate dynamic purposeful behavior in non-player characters (NPCs). In addition, well-understood AI techniques are routinely used for pathfinding. Some researchers consider NPC AI in games to be a “solved problem” for most production tasks. Games with more atypical AI include the AI director of Left 4 Dead (2008) and the neuroevolutionary training of platoons in Supreme Commander 2 (2010).[296][297]

Worldwide annual military spending on robotics rose from 5.1 billion USD in 2010 to 7.5 billion USD in 2015.[298][299] Military drones capable of autonomous action are widely considered a useful asset. In 2017, Vladimir Putin stated that “Whoever becomes the leader in (artificial intelligence) will become the ruler of the world”.[300][301] Many artificial intelligence researchers seek to distance themselves from military applications of AI.[302]

For financial statements audit, AI makes continuous audit possible. AI tools could analyze many sets of different information immediately. The potential benefit would be the overall audit risk will be reduced, the level of assurance will be increased and the time duration of audit will be reduced.[303]

A report by the Guardian newspaper in the UK in 2018 found that online gambling companies were using AI to predict the behavior of customers in order to target them with personalized promotions.[304] Developers of commercial AI platforms are also beginning to appeal more directly to casino operators, offering a range of existing and potential services to help them boost their profits and expand their customer base.[305]

Artificial Intelligence has inspired numerous creative applications including its usage to produce visual art. The exhibition “Thinking Machines: Art and Design in the Computer Age, 1959-1989” at MoMA [306] provides a good overview of the historical applications of AI for art, architecture, and design. Recent exhibitions showcasing the usage of AI to produce art include the Google-sponsored benefit and auction at the Gray Area Foundation in San Francisco, where artists experimented with the deepdream algorithm [307] and the exhibition “Unhuman: Art in the Age of AI,” which took place in Los Angeles and Frankfurt in the fall of 2017.[308][309] In the spring of 2018, the Association of Computing Machinery dedicated a special magazine issue to the subject of computers and art highlighting the role of machine learning in the arts.[310]

There are three philosophical questions related to AI:

View original post here:

Artificial intelligence – Wikipedia

Artificial Intelligence research at Microsoft

At Microsoft, researchers in artificial intelligence are harnessing the explosion of digital data and computational power with advanced algorithms to enable collaborative and natural interactions between people and machines that extend the human ability to sense, learn and understand. The research infuses computers, materials and systems with the ability to reason, communicate and perform with humanlike skill and agility.

Microsofts deep investments in the field are advancing the state of the art in machine intelligence and perception, enabling computers that understand what they see, communicate in natural language, answer complex questions and interact with their environment. In addition, the companys researchers are thought leaders on the ethics and societal impacts of intelligent technologies. The research, tools and services that result from this investment are woven into existing and new products and, at the same time, made open and accessible to the broader community in a bid to accelerate innovation, democratize AI and solve the worlds most pressing challenges.

Read the rest here:

Artificial Intelligence research at Microsoft

Benefits & Risks of Artificial Intelligence – Future of …

Many AI researchers roll their eyes when seeing this headline:Stephen Hawking warns that rise of robots may be disastrous for mankind. And as many havelost count of how many similar articles theyveseen.Typically, these articles are accompanied by an evil-looking robot carrying a weapon, and they suggest we should worry about robots rising up and killing us because theyve become conscious and/or evil.On a lighter note, such articles are actually rather impressive, because they succinctly summarize the scenario that AI researchers dontworry about. That scenario combines as many as three separate misconceptions: concern about consciousness, evil, androbots.

If you drive down the road, you have a subjective experience of colors, sounds, etc. But does a self-driving car have a subjective experience? Does it feel like anything at all to be a self-driving car?Although this mystery of consciousness is interesting in its own right, its irrelevant to AI risk. If you get struck by a driverless car, it makes no difference to you whether it subjectively feels conscious. In the same way, what will affect us humans is what superintelligent AIdoes, not how it subjectively feels.

The fear of machines turning evil is another red herring. The real worry isnt malevolence, but competence. A superintelligent AI is by definition very good at attaining its goals, whatever they may be, so we need to ensure that its goals are aligned with ours. Humans dont generally hate ants, but were more intelligent than they are so if we want to build a hydroelectric dam and theres an anthill there, too bad for the ants. The beneficial-AI movement wants to avoid placing humanity in the position of those ants.

The consciousness misconception is related to the myth that machines cant have goals.Machines can obviously have goals in the narrow sense of exhibiting goal-oriented behavior: the behavior of a heat-seeking missile is most economically explained as a goal to hit a target.If you feel threatened by a machine whose goals are misaligned with yours, then it is precisely its goals in this narrow sense that troubles you, not whether the machine is conscious and experiences a sense of purpose.If that heat-seeking missile were chasing you, you probably wouldnt exclaim: Im not worried, because machines cant have goals!

I sympathize with Rodney Brooks and other robotics pioneers who feel unfairly demonized by scaremongering tabloids,because some journalists seem obsessively fixated on robots and adorn many of their articles with evil-looking metal monsters with red shiny eyes. In fact, the main concern of the beneficial-AI movement isnt with robots but with intelligence itself: specifically, intelligence whose goals are misaligned with ours. To cause us trouble, such misaligned superhuman intelligence needs no robotic body, merely an internet connection this may enable outsmarting financial markets, out-inventing human researchers, out-manipulating human leaders, and developing weapons we cannot even understand. Even if building robots were physically impossible, a super-intelligent and super-wealthy AI could easily pay or manipulate many humans to unwittingly do its bidding.

The robot misconception is related to the myth that machines cant control humans. Intelligence enables control: humans control tigers not because we are stronger, but because we are smarter. This means that if we cede our position as smartest on our planet, its possible that we might also cede control.

See the rest here:

Benefits & Risks of Artificial Intelligence – Future of …

Robot Security Guards Will Constantly Nag Spectators at the Tokyo Olympics

Over and Over

“The security robot is patrolling. Ding-ding. Ding-ding. The security robot is patrolling. Ding-ding. Ding-ding.”

That’s what Olympic attendees will hear ad nauseam when they step onto the platforms of Tokyo’s train stations in 2020. The source: Perseusbot, a robot security guard Japanese developers unveiled to the press on Thursday.

Observe and Report

According to reporting by Kyodo News, the purpose of the AI-powered Perseusbot is to lower the burden on the stations’ staff when visitors flood Tokyo during the 2020 Olympics.

The robot is roughly 5.5 feet tall and equipped with security cameras that allow it to note suspicious behaviors, such as signs of violence breaking out or unattended packages, as it autonomous patrols the area. It can then alert security staff to the issues by sending notifications directly to their smart phones.

Prior Prepration

Just like the athletes who will head to Tokyo in 2020, Perseusbot already has a training program in the works — it’ll patrol Tokyo’s Seibu Shinjuku Station from November 26 to 30. This dry run should give the bot’s developers a chance to work out any kinks before 2020.

If all goes as hoped, the bot will be ready to annoy attendees with its incessant chant before the Olympic torch is lit. And, you know, keep everyone safe, too.

READ MORE: Robot Station Security Guard Unveiled Ahead of 2020 Tokyo Olympics [Kyodo News]

More robot security guards: Robot Security Guards Are Just the Beginning

See the article here:

Robot Security Guards Will Constantly Nag Spectators at the Tokyo Olympics

People Would Rather a Self-Driving Car Kill a Criminal Than a Dog

Snap Decisions

On first glance, a site that collects people’s opinions about whose life an autonomous car should favor doesn’t tell us anything we didn’t already know. But look closer, and you’ll catch a glimpse of humanity’s dark side.

The Moral Machine is an online survey designed by MIT researchers to gauge how the public would want an autonomous car to behave in a scenario in which someone has to die. It asks questions like: “If an autonomous car has to choose between killing a man or a woman, who should it kill? What if the woman is elderly but the man is young?”

Essentially, it’s a 21st century update on the Trolley Problem, an ethical thought experiment no doubt permanently etched into the mind of anyone who’s seen the second season of “The Good Place.”

Ethical Dilemma

The MIT team launched the Moral Machine in 2016, and more than two million people from 233 countries participated in the survey — quite a significant sample size.

On Wednesday, the researchers published the results of the experiment in the journal Nature, and they really aren’t all that surprising: Respondents value the life of a baby over all others, with a female child, male child, and pregnant woman following closely behind. Yawn.

It’s when you look at the other end of the spectrum — the characters survey respondents were least likely to “save” — that you’ll see something startling: Survey respondents would rather the autonomous car kill a human criminal than a dog.

moral machine
Image Credit: MIT

Ugly Reflection

While the team designed the survey to help shape the future of autonomous vehicles, it’s hard not to focus on this troubling valuing of a dog’s life over that of any human, criminal or not. Does this tell us something important about how society views the criminal class? Reveal that we’re all monsters when hidden behind the internet’s cloak of anonymity? Confirm that we really like dogs?

The MIT team doesn’t address any of these questions in their paper, and really, we wouldn’t expect them to — it’s their job to report the survey results, not extrapolate some deeper meaning from them. But whether the Moral Machine informs the future of autonomous vehicles or not, it’s certainly held up a mirror to humanity’s values, and we do not like the reflection we see.

READ MORE: Driverless Cars Should Spare Young People Over Old in Unavoidable Accidents, Massive Survey Finds [Motherboard]

More on the Moral Machine: MIT’s “Moral Machine” Lets You Decide Who Lives & Dies in Self-Driving Car Crashes

Link:

People Would Rather a Self-Driving Car Kill a Criminal Than a Dog


12345...102030...