Monthly Archives: April 2017

Executive Summary

As with the adoption of all technology, user experience trumps technical refinements. Many organizations implementing AI initiatives are making a mistake by focusing on smarter algorithms over compelling use cases. Use cases where peoples jobs become simpler and more productive are essential to AI workplace adoption. Focusing on clearer, crisper use cases means better and more productive relationships between machines and humans. This article offers five use case categories assistant, guide, consultant, colleague, boss that emerge when companies use AI-empowered people and processes over autonomous systems. Each describes how intelligent entities work together to get the job done and how depending on the process, AI makes the human element matter even more.

As artificial intelligence algorithms infiltrate the enterprise, organizational learning matters as much as machine learning. How should smart management teams maximize the economic value of smarter systems?

Business process redesign and better training are important, but better use cases those real-world tasks and interactions that determine everyday business outcomes offer the biggest payoffs. Privileging smarter algorithms over thoughtful use cases is the most pernicious mistake I see in current enterprise AI initiatives. Somethings wrong when optimizing process technologies take precedence over how work actually gets done.

Unless were actually automating a process that is, taking humans out of the loop AI algorithms should make peoples jobs simpler, easier, and more productive. Identifying use cases where AI adds as much value to peoples performance as to process efficiencies is essential to successful enterprise adoption. By contrast, companies committed to giving smart machines greater autonomy and control focus on governance and decision rights.

Strategically speaking, a brilliant data-driven algorithm typically matters less than thoughtful UX design. Thoughtful UX designs can better train machine learning systems to become even smarter. The most effective data scientists I know learn from use-case and UX-driven insights. At one industrial controls company, for example, the data scientists discovered that users of one of their smart systems informally used a dataset to help prioritize customer responses. That unexpected use case led to a retraining of the original algorithm.

Focusing on clearer, cleaner use cases means better and more productive relationships between AI and its humans. The division of labor becomes a source of design inspiration and exploration. The quest for better outcomes shifts from training smarter algorithms to figuring out howtheuse case should evolve. That drives machine learning and organizational learning alike.

Five dominant use case categories emerge when organizations pick AI-empowered people and processes over autonomous systems. Unsurprisingly, these categories describe how intelligent entities work together to get the job done and highlight that a personal touch still matters. Depending on the person, process, and desired outcome, AI can make the human element matter more.

Assistants

Alexa, Siri and Cortana already embody real-world use cases for AI-assistantship. In Amazons felicitous phrasing, assistants have skills enabling them to perform moderately complex tasks. Whether mediated by voice or chatbot, simple and straightforward interfaces make assistants fast and easy to use. Their effectiveness is predicated as much on people knowing exactly what they need as algorithmic sophistication. As digital assistants become smarter and more knowledgeable, their task range and repertoire expands. The most effective assistants learn to prompt their users with timely questions and key words to improve both interactions and outcomes.

Guide

Where assistants perform requested tasks, guides help users navigate task complexity to achieve desired outcomes. Using Waze to drive through cross-town traffic troubled by construction is one example; using an augmented-reality tool to diagnose and repair a mobile device or HVAC system would be another. Guides digitally show and tell their humans what their next steps should be and, should missteps occurs, suggest alternate paths to success. Guides are smart software sherpa whose domain expertise is dedicated to getting their users to desired destinations.

Consultant

In contrast to guides, consultants go well beyond navigation and destination expertise. AI consultants span use cases where workers need either just-in-time expertise or bespoke advice to solve problems. Consultants, like their human counterparts, offer options and explanations, as well as reasons and rationales. A software development project manager needs to evaluate scheduling trade-offs; AI consultants ask questions and elicit information allowing specific next step recommendations. AI consultants can include relevant links, project histories and reports for context. More sophisticated consultants offer strategic advice to complement their tactical recommendations.

Consultants customize their functional knowledge scheduling; budgeting; resource allocation; procurement; purchasing; graphic design; etc. to their human clients use case needs. They are robo-advisers dispassionately dispensing their domain expertise.

Colleague

A colleague is like a consultant but with a data-driven and analytic grasp of the local situation. That is, a colleagues domain expertise is the organization itself. Colleagues have access to the relevant workplace analytics, enterprise budgets, schedules, plans, priorities and presentations to offer organizational advice to colleagues. Colleague use cases revolve around advice managers and workers need to work more efficiently and effectively in the enterprise. An AI colleague might recommend referencing and/or attaching a presentation in an email; which project leaders to ask for advice; what budget template is appropriate for a requisition; what client contacts need an early warning, etc. Colleagues are more collaborator than tool; they offer data-driven organizational insight and awareness. Like their human counterparts, they serve as sounding boards that who? help clarify communications, aspirations and risk.

Boss

Where colleagues and consultants advise, bosses direct. Boss AI tells its humans what to do next. Boss use cases eliminate options, choices and ambiguity in favor of dictates, decrees and directives to be obeyed. Start doing this; stop doing that; change this schedule; shrink that budget; send this memo to your team.

Boss AI is designed for obedience and compliance; the human in the loop must yield to the algorithm in the system. Boss AI represents the slippery slope to autonomy the workplace counterpart to an autopilot taking over an airplane cockpit or an automotive collision avoidance system slamming on the brakes. Specific use cases and circumstances trigger human subordination to software. But bosswares true test is human: if humans arent sanctioned or fired for disobedience, then the software really isnt a boss.

As the last example illustrates, these distinct categories can swiftly blur into each other. Its easy to conceive of scenarios and use cases where guides can become assistants, assistants situationally escalate into colleagues, and consultants transform into bosses. But the fundamental differences and distinctions these five categories present should inject real rigor and discipline intoimagining their futures.

Trust is implicit in all five categories. Do workers trust their assistants to do what theyve been told or guides to get them where they want to go? Do managers trust the competence of bossware or that their colleagues wont betray them? Trust and transparency issues persist regardless of how smart AI software becomes, and they become even more important as the reasons for decisions become overwhelmingly complex and sophisticated. One risk: these artificial intelligences evolve or devolve into frenemies. That is, software that is simultaneously friend and rival to its human complement. Consequently, use cases become essential to identifying what kinds of interfaces and interactions facilitate human/machine trust.

Use cases may prove vital to empowering smart human/smart machine productivity. But reality suggests their ultimate value may come from how thoughtfully they accelerate the organizations advance to greater automation and autonomy. The true organizational impact and influence these categories may be that they prove to be the best way for humans to train their successors.

See the original post:

AI Won't Change Companies Without Great UX - Harvard Business Review

The late artist Bob Ross was known for his calm, almost ASMR-like voice, his '70s permed hair and his expert-level technique for making "happy little trees" from oil paint.

But what happens when you filter an episode of his PBS TV show "The Joy of Painting" through the neural net? You end up with a show that looks like something from a bad acid trip.

In the video "Deeply Artificial Trees" by artBoffin we see exactly what goes wrong when machine learning filters a seemingly innocent painting show into the imaginings of a sci-fi movie gone wrong.

"This artwork represents what it would be like for an AI to watch Bob Ross on LSD (once someone invents digital drugs)," artBoffin writes in the video description. "It shows some of the unreasonable effectiveness and strange inner workings of deep learning systems. The unique characteristics of the human voice are learned and generated, as well as hallucinations of a system trying to find images which are not there."

At the beginning of the video, Ross pets what looks like a gerbil from hell. The painting Ross is working on should have happy little trees, but instead it's infested with giant cockroaches. And it just gets weirder from there.

Watching the video I saw numerous horrific, squiggly animal creations that would inspire the likes of H.P. Lovecraft. I may never be able to fall sleep again.

Batteries Not Included: The CNET team shares experiences that remind us why tech stuff is

Solving for XX: The industry seeks to overcome outdated ideas about "women in tech."

See more here:

Bob Ross painting trees via AI is like a drug-fueled nightmare - CNET

The good people at Facebook must be working overtime.

After implementing the Snapchat-esque Stories feature and trialing a second News Feed, the company is adding some AI components to its Messenger's M assistant.

Launching in the US on Thursday, Facebook said in a blog that M will offer helpful suggestions during conversations with friends.

M knows if you're talking about buying something off a friend, for instance, and will automatically offer payment options. Facebook also said M can offer to share your exact location with a friend during a conversation, and will offer the option of a poll if you're in a group chat and something needs deciding.

M's AI abilities have hit both iOS and Android, but right now they're only available to users in the US. It was noted, though, they will "eventually roll out to other countries." Facebook also promised this was the beginning, saying M's predictive powers will only get bigger from here.

Tech Enabled: CNET chronicles tech's role in providing new kinds of accessibility.

Technically Literate: Original works of short fiction with unique perspectives on tech, exclusively on CNET.

Read the rest here:

Facebook Messenger's M assistant gets new AI powers - CNET

The video includes some tools we already knew about -- mainly the ability to copy one photo's style and look to another in a couple of taps. Adobe researchers worked with Cornell University to employ AI to take things like color, lighting and contrast you really like in one image and apply it to a boring ol' crappy photo. While that tool is part of an experimental app called "Deep Photo Style Transfer" that's posted on Github, it looks like Adobe has plans to bring that feature to a more robust piece of mobile software.

Thanks to Adobe Sensei, a mobile app could also allow for easy perspective editing and automatic photo masking. A liquify tool updates to the perspective of a selfie with a slider, keeping the subject's face in proportion while the edits are applied. A similar tool has been available inside Photoshop Fix for a while now, but the so-called Face-Aware version just hit Photoshop on the desktop last summer. If you need to adjust the depth of field, portrait masking can help you easily do that with a simple slider adjustment. Adobe hasn't been shy about bringing desktop-friendly features to mobile, so don't be surprised if this masking feature makes the leap.

While all of these tools make for a compelling photo-editing app, there's no indication when (or if) Adobe will put them in a piece of software you can actually use. Given its recent mobile focus, you can bet more powerful features are coming to the likes of Photoshop Fix and other apps. It's only a matter of time.

Read more from the original source:

Adobe shows how AI can work wonders on your selfie game - Engadget

Forbes

Blue-Collar Revenge: The Rise Of AI Will Create A New Professional Class Forbes New, more-modern manufacturing processes, including the use of robots, have gutted the number of high-paying factory jobs in the U.S. and caused economic angst in large portions of the country. The movement of manufacturing plants overseas has ...

Originally posted here:

Blue-Collar Revenge: The Rise Of AI Will Create A New Professional Class - Forbes

Executive Summary

Curiosity has been hailed as one of the most critical competencies for the modern workplace. As the workplace becomes more and more automated, it begs the question: Can artificial intelligence ever be curious as human beings? AIs desire to learn a directed task cannot be overstated. Most AI problems comprise defining an objective or goal that becomes the computers number one priority.At the same time, AI is also constrained in what it can learn. AI is increasinglybecoming a substitute for tasks that once required a great deal of human curiosity, and when it comes to performance, AI will have an edge over humans in a growing number of tasks. But the capacity to remain capriciously curious about anything, including random things, and pursue ones interest with passion, may remain exclusively human.

Curiosity has been hailed as one of the most critical competencies for the modern workplace. Its been shown to boost peoples employability. Countries with higher curiosity enjoy more economic and political freedom, as well as higher GDPs. It is therefore not surprising that, as future jobs become less predictable, a growing number of organizations will hire individuals based on what they could learn, rather than on what they already know.

Of course, peoples careers are still largely dependent on their academic achievements, which are (at least partly) a result of their curiosity. Since no skill can be learned without a minimum level of interest, curiosity may be considered one of the critical foundations of talent. AsAlbert Einstein famously noted,I have no special talent. I am only passionately curious.

How it will impact business, industry, and society.

Curiosity is only made more important for peoples careers by the growing automation of jobs. At this years World Economic Forum, ManpowerGroup predicted that learnability, the desire to adapt ones skill set to remain employable throughout ones working life, is a key antidote to automation. Those who are more willing and able to upskill and develop new expertise are less likely to be automated. In other words, the wider the range of skills and abilities you acquire, the more relevant you will remain in the workplace. Conversely, if youre focused on optimizing your performance, your job will eventually consist of repetitive and standardized actions that could be better executed by a machine.

But what if AI were capable of being curious?

As a matter of fact, AIs desire to learn a directed task cannot be overstated. Most AI problems comprise defining an objective or goal that becomes the computers number one priority. To appreciate the force of this motivation, just imagine if your desire to learn something ranked highest among all your motivational priorities, above any social status or even your physiological needs. In that sense, AI is way more obsessed with learning than humans are.

At the same time, AI is constrained in what it can learn. Its focus and scope are very narrow compared to that of a human, and its insatiable learning appetite applies only to extrinsic directives learn X, Y, or Z. This is in stark contrast to AIs inability to self-direct or be intrinsically curious. In that sense, artificial curiosity is the exact opposite of human curiosity; people are rarely curious about something because they are told to be. Yet this is arguably the biggest downside to human curiosity: It is free-flowing and capricious, so we cannot boost it at will, either in ourselves or in others.

To some degree, most of the complex tasks that AI has automated have exposed the limited potential of human curiosity vis-a-vis targeted learning. In fact, even if we dont like to describe AI learning in terms of curiosity, it is clear that AI is increasingly a substitute for tasks that once required a great deal of human curiosity. Consider the curiosity that went into automobile safety innovation, for example. Remember automobile crash tests? Thanks to the dramatic increase in computing power, a car crash can now be simulated bya computer. In the past, innovative ideas required curiosity, followed by design and testing in a lab. Today, computers can assist curiosity efforts by searching for design optimizations on their own. With this intelligent design process, the computer owns the entire life cycle of idea creation, testing, and validation. The final designs, if given enough flexibility, can often surpass whats humanly possible.

Similar AI design processes are becoming more common across many different industries. Google has used it to optimize cooling efficiency with itsdata centers. NASA engineers have used it to improve antennae quality for maximum sensitivity. With AI, the process of design-test-feedback can happen in milliseconds instead of weeks. In the future, the tunable design parameters and speed will only increase, thus broadening our possible applications for human-inspired design.

A more familiar example might be the face-to-face interview, since nearly every working adult has had to endure one. Improving the quality of hires is a constant goal for companies, but how do you do it? A human recruiters curiosity could inspire them to vary future interviews by question or duration. In this case, the process for testing new questions and grading criteria is limited by the number of candidates and observations. In some cases, a company may lack the applicant volume to do any meaningful studies to perfect itsinterview process. But machine learning can be applied directly to recorded video interviews, and the learning-feedback process can be tested in seconds. Candidates can be compared based on features related to speech and social behavior. Microcompetencies that matter such as attention, friendliness, and achievement-based language can be tested and validated from video, audio, and language in minutes, while controlling for irrelevant variables and eliminating the effects of unconscious (and conscious) biases. In contrast, human interviewers are often not curious enough to ask candidates important questions or they are curious about the wrong things, so they end up paying attention to irrelevant factors and making unfair decisions.

Lastly, consider a human playing a computer game. Many games start out with repeated trial and error, sohumans must attempt new things and innovate to succeed in the game: If I try this, then what? What if I go here? Early versions of game robots were not very capable because they were using the full game state information; they knew where their human rivals were and what they were doing. But since 2015something new has happened: Computers can beat us on equal grounds, without any game state information, thanks to deep learning. Both humans and the computers can make real-time decisions about their next move. (As an example, see this video of a deep network learning to play the game Super Mario World.)

From the above examples, it may seem that computers have surpassed humans when it comes to specific (task-related) curiosity. It is clear that computers can constantly learn and test ideas faster than we can, so long as they have a clear set of instructions and a clearly defined goal. However, computers still lack the ability to venture into new problem domains and connect analogous problems, perhaps because of their inability to relate unrelated experiences. For instance, the hiring algorithms cant play checkers, and the car design algorithms cant play computer games. In short, when it comes to performance, AI will have an edge over humans in a growing number of tasks, but the capacity to remain capriciously curious about anything, including random things, and pursue ones interest with passion may remain exclusively human.

Read this article:

Can AI Ever Be as Curious as Humans? - Harvard Business Review

Ever heard of Chainer, the open-source framework for creating neural networks?

I hadn't either until yesterday Intel decided to give it a big hug, taking Chainer from being big in Japan, where its parent company Preferred Networks works with the likes of Toyota on secret projects, to rather greater prominence.

Chainer can use the help: launched in 2015 and open-sourced last year, the tool's GitHub repo is busy but hardly the most lively place on the internet.

That's probably about to change because Intel has decided Chainer is a fine way to develop AI workloads that create demand for its silicon. Doubly so if it can be taught to speak fluent Xeon, instead of only chatting to NVIDIA GPUs as was previously the case. The deal between Intel and Preferred means Chainer will from now on be developed for Intel architectures and changes shared on Intel's GitHub repo for the project.

Why should we care that Intel's decided to give Chainer a leg-up?

On the purely technical side of things, it looks like good gear. Chainer CEO Toru Nishikawa yesterday showed Intel's AI Day in Tokyo the slide below on which he claims to have made Google's TensorFlow look like it was working in treacle when measured on training time for image net classifications. Nishikawa-San also said Chainer had tied in a recent Amazon.com test to train robots to pick stock.

So you could do worse than have a look if you are thinking about neural networks.

But the Chainer tie-up is also worth considering because it shows how Intel builds markets and will try to make itself the dominant player in Artificial Intelligence, a market widely assumed to be on the cusp of a boom.

It's also a market that is currently keen on GPUs. So Intel wants to build a portfolio of products to make Xeons the heart of AI, not GPUs.

Intel's not, however, using SciFi definitions of AI. Amir Khosrowshahi, former CTO of Nervana and now holder of the same position in Intel's new AI Group, prefers to describe AI as involving deep statistical analysis of very closely-observed events so that we can infer likely outcomes with satisfying precision.

Modern hardware can do that analysis and wrangle the necessary mountains of data collected to make the analysis useful, but it mostly brute-forces it. Dedicated hardware will speed things up and that's where Intel is going, by building and/or buying that hardware and building the software ecosystem to match.

You may have seen this movie before when virtualization was obviously the next big thing and Intel added extensions to its silicon so it would be especially good at hosting multiple VMs. Chipzilla's also done things like bridge the Lustre and HDFS file systems so that HPC clusters runnings Lustre could run Hadoop, which relies on HDFS. Intel wins either way: it's invested in Hadoop provider Cloudera and has lots of HPC customers who didn't scream when Chipzilla made their rigs more useful. Intel's also optimised its consumer CPUs for video transcoding because editing HD home movies without having to stay up all night is one of the few compelling reasons to buy a new PC.

Intel's now using the same playbook for AI. Buying field-programmable gate array (FPGA) vendor Altera gave Intel the tech to build hybrid Xeons that offer integrated programmability so you can get silicon speed for exotic analyses that would make a vanilla Xeon weep. Altera is now working to make sure that developing code for FPGAs, once the province of embedded systems engineers, is not a stretch for the average Java developer.

Bernhard Friebe, Intel's director of planning and marketing for FPGA Design Software and Intellectual Property, said Intel is developing libraries for common AI tasks, gives them away and has built tools that mean developers need to write just one line of code to target FPGAs.

Nervana gave Intel silicon tailored to AI and lots of the software developers will need to use it.

The two companies also give Intel the technology it will one day bake into Xeons so they become better at the kind of data-crunching AI needs.

We'll see those products emerge later in 2017 when the Lake Crest Xeon adds a discrete AI accelerators for AI workloads. The Skylake Xeon with a joined-at-the-hip FPGA, code-named Knights Crest, will debut later in the same year. Intel's being shy about exact specs, but Both use proprietary inter-chip links and a new architecture called Flexpoint to improve parallelism. But they're early products: both promise 10x parallelism. By 2020 Intel pledges to reduce the time needed to train an AI model by a factor of 100.

But the main game here is that by adding AI abilities to those Xeons, Intel can talk to mainstream users about doing AI with familiar kit, rather than wrapping their head around GPUs. And it can point to Chainer and many other software investments to show that existing developers won't struggle to at least start playing with AI.

The company still has exotica up its sleeve. Barry Davis, general manager of Intel's Accelerator Workload Group, told El Reg that by the second half of 2018 we'll also see Knight's Mill, the next-generation Xeon Phi optimised for AI. Details of the product are scarce, but Intel is talking up the fact it will be able to address up to 400GB of memory, far more than some GPUs.

Once everyday Xeons are good at AI, there will be little excuse not to consider them. More exotic products like Xeon Phi or FPGA-bonded Xeons can also run in the cloud, where users can try them out without capital expenditure.

By the time the ready-for-AI range is mature, Chainer will have been running on Intel hardware for about three years, will probably be rather improved thanks to the input Intel's support will have generated.

That won't tup tip anyone over into a decision to go Intel when contemplating AI. But bringing Chainer into Intel's world is one of a dozen or a hundred other efforts. Some of those efforts are blindingly obvious billion-dollar acquisitions. Some are imperceptible nudges to useful open source projects. Others will be thoroughly obscure instructions issued to server-makers.

They'll all add up to an ecosystem designed to make Intel all-but-impossible to leave off a list of vendors to consider when doing AI .

Of course the world's not going to stand still and let Intel do this. But Chipzilla is confident it can dominate any rivals.

At this point it's tempting to point out that Intel is nowhere in mobile, a field in which it felt its modus operandi would work but ended up being slaughtered by ARM.

Barry Davis thinks Intel has figured out why: his version of recent history says ARM always wanted to start at the edge of the network and work its way in to the data centre. In the mobile field, Intel tried to work the way it had with PCs but found itself surrounded, late to the party and without the right friends once it arrived. The execs I met yesterday didn't dismiss challenges ARM presents in AI, but feel that as ARM is yet to become a significant data centre player and therefore isn't in a position to spearhead an ecosystem-creating challenge that will satisfy businesses and developers.

Of course Intel would say that, wouldn't it? Or is its confidence derived from deep statistical analysis of closely-observed events that let it infer likely outcomes with satisfying precision?

Continue reading here:

Big-in-Japan AI code 'Chainer' shows how Intel will gun for GPUs - The Register

Image: iStock/Getty Images

The American Heart Association published the results of a trial that shows stroke survivors are twice as likely to take anti-blood clot treatments when they are using an artificial intelligence (AI) platform, compared to those receiving more traditional treatment.

The AI platform, AiCure, uses software algorithms on smartphones to confirm patient identify, the medication, and if the medication was taken. Patients receive automated reminders and dosing instructions as well. Healthcare workers receive real-time data which allows for early detection of patients who are not taking their meds as scheduled.

SEE: Google's DeepMind and the NHS: A glimpse of what AI means for the future of healthcare (ZDNet)

This latest trial, which lasted 12 weeks and was published in the American Heart Association's journal Stroke, shows more of AI's potential. Anti-blood clot medication can prevent another stroke, so it is essential that patients take their medication. Approximately 800,000 people suffer a stroke annually and it is the fifth leading cause of death.

"Many patients are unable to self-manage and are at increased risk of stroke and bleeding. The use of technology and artificial intelligence has the potential to significantly improve health outcomes and reduce costs in clinical care," said Laura Shafner, study coauthor and chief strategy officer at AiCure, in a press release.

AI has vast potential in the healthcare industry. It reduces the tasks that medical professionals must perform and that saves organizations money. Plus, there is plenty of data that healthcare generates and AI systems can be trained to take advantage of this and provide useable information to healthcare providers.

IBM Watson is also busy working on various AI tools for healthcare, such as a chip that can diagnose a potentially fatal condition, a camera that can scan a pill to see if it is real or counterfeit, and a system to identify mental illness. And there are others. An AI program from Behold.ai helps doctors identify cancer and medical abnormalities. Also an AI app developed by the University of Rochester tracks foodbourne illness and helps public health departments spot public health outbreaks. As AI becomes more commonplace, more options will exist to help patients with their health.

Also see:

Follow this link:

New study shows how AI can improve recovery in stroke patients - TechRepublic