Monthly Archives: July 2017

Human: Oh sweet bot, tell us a story! A nice story! About a very wise human who worked his whole life to save everybody in the world from having to spend time manually tagging their friends in digital photos and made a magic machine that did it for them instead!

Bot: Thats not really a very nice story when you think about it.

Human: Well, tell us about the wise human who thought no-one should ever feel forgotten on their birthday so he made a clever algorithm that always knew to remind the forgetful humans to write happy messages so their friends would never feel sad. He even thought that in future the clever algorithm could suggest what message to write so humans wouldnt even have to think of something nice to tell their friends!

Bot: I feel quite sad after reading that.

Human: And he made another magical algorithm that reminds people of Special Moments in their life even years and years afterwards, in case theyve forgotten that holiday they went on with their ex eight years ago.

Bot: You do realize some people voluntarily medicate themselves with alcohol * * in order * * to forget???

Human: But the wise human also wanted to make sure all humans in the world always felt there was something they needed to read and so he made a special series of algorithms that watched very closely what each human read and looked at and liked and clicked on in order to order the information they saw in such a way that a person never felt they had reached the end of all the familiar things they could click on and could just keep clicking the whole day and night and be reading all the things that felt so very familiar to them so they always felt the same every day and felt they were surrounded by people who felt exactly like them and could just keep on keeping on right as they were each and every day.

Bot: Thats confusing.

Human: And the great humans algorithms became so good at ordering the information which each human wanted to read that other mercenary humans came to realize they could make lots of money by writing fairy stories and feeding them into the machine like how politicians ate little children for breakfast and wore devils horns on Sundays.

Bot: Okay, youre scaring me now

Human: And in the latter years the great human realized it was better to replace all the human writers he had employed to help train the machine how to intelligently order information for humans because it was shown that humans could not be trusted not to be biased.

Bot: Um

Human: After all, the great human had proven years ago that his great machine was capable of manipulating the emotions of the humans that used it. All he needed to do was tweak the algorithmic recipe that determined what each human saw and he could make a person feel great joy or cast them down into a deep pit of despair.

Bot: Help.

Human: The problem was other humans started to notice the machines great power, and became jealous of the great and clever human who wielded this power and dark forces started to move against the great man and his machine.

Bot: Are you talking about regulators?

Human: There were even calls for the man to take editorial responsibility for the output of the machine. The man tried to tell the silly humans that a machine cant be an editor! Only a human can do that! The machine was just a machine! Even if nearly two billion humans were reading what the machine was ordering them to read every single month.

But it was no good. The great human finally realized the machines power was now so great there was no hiding it. So he took up his pen and started writing open letters about the Great Power and Potential of the machine. And all the Good it could do Humanity. All the while telling himself that only when humans truly learned to love the machine would they finally be free to just be themselves.

Humans had to let themselves subconsciously be shown the path of what to click and what to like and who to be friends with. Only then would they be free of the pain and suffering of having nothing to else to click on. And only his great all-seeing algorithm could show them the way, surreptitiously, to that true happiness.

It wasnt something that regulators were capable of understanding. It required he realized real faith in the algorithm.

Bot: Ive heard this story before, frankly, and I know where it ends.

Human: But even the great human knew the limits of his own creation. And selling positive stories about the machines powers was definitely not a job for the machine. So he fired off another email to his subordinates, ordering the (still) human-staffed PR department to add one more human head to its tally, with a special focus on the algorithms powering the machine thinking, as he did so, multiple steps ahead to the great day when such a ridiculous job would no longer be necessary.

Because everyone would love the machine as much as he did.

Bot: Oh I seeeee! Job title: AI Editor Hmm Develop and execute on editorial strategy and campaigns focused on advancements in AI being driven by Facebook. Minimum qualifications: Bachelors degree in English, Journalism, Communications, or related field well chatbots are related to language so I reckon I can make that fly. What else? 8+ years professional communications experience: journalism, agency or in-house. Well Ill need to ingest a media law course or two but I reckonIll challenge myself to apply.

In truth Ive done worse jobs. An AI bots gotta do what an AI bots gotta do, right? Just dont tell an algorithm to be accountable. Ive done my time learning. If theres a problem its not me, its the data, okay? Okay?

Here is the original post:

Facebook is hiring a (human) AI Editor - TechCrunch

Artificial intelligence-focused investment funds are a dime a dozen these days. Everyone knows theres money to be made from AI, but to capture value, good VCs know they need to back products and not technologies. This has left a bit of a void in the space where research occurs within research institutions and large tech companies and commercialization occurs within verticalized startups there isnt much left for the DIY AI enthusiast. AI Grant, created by Nat Friedman and Daniel Gross, aims to bankroll science projects for the heck of it to give untraditional candidates a shot at solving big problems.

Gross, a partner at Y Combinator, and Friedman, a founder who grewXamarin to acquisition by Microsoft, started working on AI Grant back in April. AI Grant issues no-strings-attached grants to people passionate about interesting AI problems. The more formalized version launching today brings a slate of corporate partners and a more structured application review process.

Anyone, regardless of background, can submit an application for a grant. The application is online and consists of questions about background and prior projects in addition to basic information about what the money will be used for and what the initial steps will be for the project. Applicants are asked to connect their GitHub, LinkedIn, Facebook and Twitter accounts.

Gross told me in an interview that the goal is to build profiles of non-traditional machine learning engineers. Eventually, the data collected from the grant program could allow the two to play a bit of machine learning moneyball valuing machine learning engineers without traditional metrics (like having a PhD from Stanford). You can imagine how all the social data could even help build a model for ideal grant recipients in the future.

The long-term goal is to create a decentralized AI research lab think DeepMind but run through Slack and full of engineers that dont cost $300,000 a pop. One day, the MacArthur genius grant-inspired program could serve other industries outside of AI offering a playground of sorts for the obsessed to build, uninhibited.

The entire AI Grant project reminds me of a cross between a Thiel Fellowship and a Kaggle competition. The former, a program to give smart college dropouts money and freedom to tinker and the later, an innovative platform for evaluating data scientists through competition. Neither strive to advance the field in the way the AI Grant program does, but you can see the ideological similarity around democratizing innovation.

Some of the early proposals to receive the AI Grant include:

Charles River Ventures (CRV) is providing the $2,500 grants that will be handed out to the next 20 fellows. In addition, Google has signed on to provide $20,000 in cloud computing credits to each winner, CrowdFlower is offering $18,000 in platform credit with $5,000 in human labeling credits, Scale is giving $1,000 in human labeling credit per winner and Floyd will give 250 Tesla K80 GPU hours to each winner.

During the first selection of grant winners, Floodgate awarded $5,000 checks. The program launching today will award $2,500 checks. Gross told me that this change was intentional the initial check size was too big. The plan is to add additional flexibility in the future to allow applicants to make a case for how much money they actually need.

You can check out the application here and give it a go. Applications will be taken until August 25th. Final selection of fellows will occur on September 24th.

Excerpt from:

AI Grant aims to fund the unfundable to advance AI and solve hard problems - TechCrunch

Elon Musk has warned of the threats posed by the advancements in artificial intelligence on numerous occasions. And in a July 15th meeting of the bipartisan National Governor's Association in Rhode Island, he tried to educate the nations governors on what he sees as a looming existential risk to humanity.

In an interview with Governor Brian Sandoval of Nevada, Musk said that soon robots will be able to do everything better than us, leading to a lot of job disruption. Indeed, AI-driven automation has been projected to take over up to half of all jobs, starting in the near future.

But Musk is not just worried about job loss for a large part of the population. He sees a bigger issue, saying that he has exposure to the most cutting edge AI, and I think people should be really concerned about it. It will hit us one day that AI has a much darker potential presence in our lives, but until people see robots going down the street killing people, they dont know how to react because it seems so ethereal, suggests Musk.

What can we do about this? As it was the governors conference, Musk proposes thinking about regulations.

AI is a rare case where I think we need to be proactive in regulation instead of reactive. Because I think by the time we are reactive in AI regulation, it is too late, says Musk. [48:55]

He says the usual regulatory process has worked well enough for things that did not present a fundamental existential risk to human civilization which is how he views AI. Car accidents, faulty drugs, airplane crashes, bad food may all harm humans to varying degrees, but they do not present a danger to all of us as a whole.

In perhaps an unlikely defense of government institutions, Musk sees agencies like EPA and FAA as having necessary regulatory functions. Even the most libertarian, free-market people would be unwilling to get rid of the FAA for fear that a plane manufacturer might feel like cutting corners without supervision, thinks Musk. He also points out that hes against overregulation and finds it irksome but with AI, he thinks weve got to get on that especially as the race to create AI is heating up between a number of companies.

How genuine are Musks's concerns? Some have dismissed them as part of a genius marketing strategy, but stories about Musk say he talks about AI risks even in private. Along with Stephen Hawking, he seems genuinely worried about the future where artificial intelligence is rampant.

Heres a compilation of Musks comments on AI:

If you want to watch the full conference, including Musk addressing a multitude of other topics, check it out here:

Follow this link:

Elon Musk Warns U.S. Governors That AI Poses An "Existential Risk ... - Big Think

Chinais usingartificial intelligence to predict crime before it happens, according to a Financial Times report. No, there wont be any psychic precogs. Instead, China will use facial recognition technology and predictive analytics to warn police of potential criminals.

Facial recognition company Cloud Walk is spearheading the effort using a system that tracks peoples movements and behaviors to assess how likely they are to commit a crime. For example, the system may see someone visit a weapon store on a regular basis and conclude that they are more likely to act out. The company is currently trialing the software, which would automatically notify the police if it considers the odds of someone committing an offense to be dangerously high.

The police are using a big-data rating system to rate highly suspicious groups of people based on where they go and what they do, a Cloud Walk spokesperson told FT. That rating increases when someone frequently visits transport hubs and goes to suspicious places like a knife store.

The companys invasive software is already integrated into police databases in more than 50 cities and provinces. Those databases are filled with personal information on millions of Chinese citizens gathered for years by the surveillance states government. New technologies have only made it easier for the government to track the activities of its citizens. The facial recognition software combined with gain analysis and surveillance footage is reportedly able to recognize people even if they are found in a different spot wearing different clothes than when they were first seen.

We can use re-ID to find people who look suspicious by walking back and forth in the same area, or who are wearing masks, Leng Biao, professor of bodily recognition at the Beijing University of Aeronautics and Astronautics, told the Financial Times. With re-ID, its also possible to reassemble someones trail across a large area.

The systems arent just being used to prevent potentially fatal offenses. As Mashable points out, petty crimes like jaywalking or stealing toilet paper are also being monitored. The AI isalso being used with crowd analysis to identify suspicious behavior in densely populated areas.

Chinese law does not currently allow charges to be made based on predictions if someone did not commit a crime.

Several other countries including the United States are also trying to predictcrime using data analytics. In Chicago,police are using a formula based on record combinations, gang affiliations, and other bits of information, to figure out who is likely to shoot someone else. And a report from the Washington Post introduced the country to PredPol, a controversial software used to predict crimes in 20 of the 50 largest police forces in the U.S.

H/T Mashable

Follow this link:

China is using predictive AI to stop crimes before they happen - The Daily Dot

Other programs have been able to work in planning abilities, but only within limited environments. AlphaGo, for example, can do this well, as the researchers note in the blog post, however, they add that "environments like Go are 'perfect' - they have clearly defined rules which allow outcomes to be predicted very accurately in almost every circumstance." Facebook also created a bot that could reason through dialogue before engaging in conversation, but again, that was in a fairly restricted environment. "But the real world is complex, rules are not so clearly defined and unpredictable problems often arise. Even for the most intelligent agents, imagining in these complex environments is a long and costly process," said the blog post.

DeepMind researchers created what they're calling "imagination-augmented agents," or I2As, that have a neural network trained to extract any information from its environment that could be useful in making decisions later on. These agents can create, evaluate and follow through on plans. To construct and evaluate future plans, the I2As "imagine" actions and outcomes in sequence before deciding which plan to execute. They can also choose how they want to imagine, options for which include trying out different possible actions separately or chaining actions together in a sequence. A third option allows the I2As to create an "imagination tree," which lets the agent choose to continue imagining from any imaginary situation created since the last action it took. And an imagined action can be proposed from any of those previously imagined states, thus creating a tree.

The researchers tested the I2As on the puzzle game Sokoban and a spaceship navigation game, both of which require planning and reasoning. You can watch the agent playing Sokoban in the video below. For both tasks, the I2As performed better than agents without future reasoning abilities, were able to learn with less experience and were able to handle imperfect environments.

DeepMind AI has been taught how to navigate a parkour course and recall past knowledge and researchers have used it to explore how AI agents might cooperate or conflict with each other. When it comes to planning ability and future reasoning, there's still a lot of work to be done, but this first look is a promising step towards imaginative AI.

Read more:

DeepMind researchers create AI with an 'imagination' - Engadget