Membership Signup
Singularity University

Facebook Building Major Artificial Intelligence System To Understand Who We Are

Meaningful artificial intelligence has been the aim of computer science since Alan Turing first imagined, in the 1950s, a computer that “passed” as human. Hollywood movies began shortly thereafter to depict computers with human-like intelligence. But, like so many things, artificial intelligence has been much harder to achieve in reality than in the movies.

But following research breakthroughs about five years ago, leading tech companies, including Microsoft, IBM and Google, have begun investing big money in artificial intelligence applications.

Facebook, not wanting to be left behind, is flexing its muscles, too. In addition to expanding its use of facial recognition, the company is developing deep learning software to understand what its users say and do online, a spokesman confirmed for Singularity Hub. (The project was first reported in MIT’s Technology Review.)

Much of Wall Street’s interest in Facebook, albeit fickle, has stemmed from the potential commercial value of what the social network knows about its users. Unless Facebook is considering major changes to its business plan, the company will use artificial intelligence to learn more about its users than what they share.

And Facebook has a lot of data from which an artificial learning setup could draw inferences.

“Places like Google and Facebook now have so much data that the kinds of networks you can train and the kinds of things you can learn from that have basically been unimaginable,” Bhiksha Raj, a machine learning expert at Carnegie Mellon University, told Singularity Hub.

“If you feed all this information, the computer is basically going to develop the ability to make sense of the inputs. Then you’ve trained a network that can scan your page and those of your friends and tell you all kinds of things about you that you hadn’t put out there, and that really is the holy grail for all of these companies,” Raj said.

It’s possible, too, that the agreements that govern what Facebook does with users’ data won’t apply to the data the company generates about its users with its neural networks.

“It’s kind of no man’s land at this point. We haven’t gotten to the point where people have begun worrying about that kind of thing, but we’re going to get there pretty fast,” Raj said.

Notable deep learning projects have already allowed computers to recognize in photos and videos the faces of humans and cats and to identify the emotions behind written content even when they’re not stated explicitly. And before Facebook acquired the company in 2012, Face.com claimed its software could identify which photographed smiles were genuine and which were fake.

AI-agsandrewFacebook has more experience with facial recognition applications of artificial intelligence, but it moved in to natural language processing with the search feature it launched this summer.

Facebook’s current staffing for the AI project suggests the company will use deep learning to dive deeper into both computer vision and natural language processing. Face.com co-founder Yaniv Taigman and Lubomir Bourdev will contribute expertise on computer vision. But natural language expert Srini Narayanan is building the project’s staff, and Graph Search veteran Keith Adams will also move to the AI project.

Facebook has poached Marc’Aurelio Ranzato from Google’s deep learning project, likely to deliver expertise in how to start the computers learning. Ranzato studied under Geoffrey Hinton, the University of Toronto computer scientist whose research into how best to train deep neural networks has spurred much of the current activity in the field.

Deep learning builds on the longtime use of “neural networks,” or computing systems that simulate the teamwork among neurons in the human brain. Instead of having each neuron, or detector, look for a particular input, which requires too many detectors to be plausible, scientists assign each layer a different task that builds on the first.

“You’re attempting to [have the computer] learn a succession of levels of representation of increasing complexity and abstraction,” Christopher Manning explained in an influential video tutorial on deep learning.

Once the layers have been set up, they can be fine-tuned to handle different sorts of inputs. In other words, such a system will be a big investment for Facebook but one that promises big payoffs.


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>


  • Calum says:

    Facebook is probably not setting out to create a conscious machine – a human-level AI – but if you combined the machine learning capabilities of Facebook, Google, IBM, and Microsoft you would go a long way in that direction. And Google and IBM actually ARE trying to build conscious machines.
    This is brain-building from the bottom up: replicating component parts of human intelligence. Other organisations, like Henry Markram’s Human Brain project and Obama’s BRAIN project, are spending similarly large sums of money on a top-down approach: copying the structure and activity of human brains.
    A lot of AI researchers think that one or other of these approaches will be successful by the middle of this century. And if they are, then what? If that question interests you, take a look at http://bit.ly/170uDbM

  • dobermanmacleod says:

    Facebook in some ways has more intimate data on their users than Google. The use of AI in terms of evaluating user’s psychology and personal economics makes a whole lot of sense for Facebook’s business model. Such refined data ought to be very very valuable, some would say invaluable. Those proprietary algorithms that Facebook develops will also be invaluable. BTW, Facebook ought to enter into a partnership with Google, rather than reinventing the wheel.

    • Calum says:

      You may be right that Facebook has more intimate data on its users than Google, although I imagine many people divulge more about themselves in emails than they do on Facebook. A secret, after all, is something you only tell one person at a time.
      But the real difference, it seems to me, is that Facebook’s main driving mission is to connect people and get them sharing. AI is a means to an end for them. One of Larry Page’s main goals is to build a mind, which is why he hired Kurzweil.
      However, what really matters is not which company contributes most to the project of making a conscious machine, but whether it is do-able, how soon, and will we like the result?

      • Ryan Howe says:

        I do not disagree that both Google and Facebook have very potentially valuable data, however secrets spread with either will be equally salacious. People chat and message with Facebook in the web and mobile apps and that can lead to the kind of intimate psychological data you may hinting at as. Of course, it is much easier and common to make dummy accounts for passing very sensitive information.

        Nonetheless, it is something premature even to contemplate setting up ML with such breadth that it would be able to parse enough data to begin determining that kind of nuance. I am by no means in anyways saying it won’t happen or can’t, just I prefer to think in the 3-8 time for some of these projects to bear fruit.

        Additionally I had once been of the mind that if only IBM, Google and Facebook could partner, IBM more so for the hardware, namely the work on neuromorphic chips but also their history. I can’t however see it ever happening though, if they include Apple in then it would be quashed as for anti-competitiveness because it is. Regardless of their combined capital being somewhere near half a trillion, the business proposal is unorthodox as the nature of the cooperation is so vast and needs to be justified as business I don’t see a 3-way custody negotiation over the most advanced AI application in history being amicable…it’d more or less balkanize them and guarantee a later conglomeration. As pathetically underwhelming corporate legislation and enforcement is, I don’t really see any country being able to look past what it would mean for them. The GOP can crusade on shutting down the federal work force but if you noticed around a month and a half ago a temporary black out of Google took out 40% of internet traffic.

      • Cameron Scott
        Cameron Scott says:

        I have to disagree with part of what you say, Calum. Both Google and Facebook share the same business model: Sell more ads for more money. Sharing, emailing and even AI are just the means to that end.

        • Calum says:

          Hi Cameron. AI is indeed invaluable to both companies, and I’m sure you’re right that at a high enough level of abstraction, their business models are the same. But that’s a bit like saying that Amazon and Exxon have the same business model because they both make money by selling us products.

          But my point is not to do with their business models. As far as I’m aware, Mark Zuckerberg has never said that he wants to build an AGI, a conscious machine, for its own sake. Larry Page has said that, and it’s why he hired Ray Kurzweil.

          I’m not placing any bets on whether the first AGI will be created by Google, Facebook, IBM, DARPA, the Human Brain Project in Lausanne, Paul Allen or somebody else – because I simply don’t know. My concerns are: a) is the thing possible in the first place, and if so, b) when, and c) what impact will it have on us? It seems to me that the answers are: a) probably, b) soon, and c) either wonderful or disastrous.

          Given all that, I’m keen to see the issues debated more widely.

          • Ryan Howe says:

            Calum, I was reticent initially when I saw your questions as a I usually am in those areas, not for their merit (they are the core, and really all that matters) but that it has been played and replayed so many times it can become difficult to approach a discussion with the same eagerness.

            a) It is quite clear that AGI is possible as we are the proof, despite the circular reasoning one may run into then.
            b) The question really is specificity, as where can we expect a disappoint in our projection. I will note Kurzweil as he is so easy a mark to look at in this respect. For instance,


            As you can see, he marks failure in many cases because he expects a very specific and exact response. I was of this persuasion for a while, but I have since changed as with technology and most predictions therein if go in expecting perfection you will come out with nothing, why? The world is vastly complex and tolerances less than say 3-4 years could kick your ass there.
            c) It’d the be the single most historic and important moment in human history. However, I cannot in good faith really honestly comment as I believe that what you, or many others, see as disastrous is more transformative and that it isn’t really wonderful but more so beautiful, alien and destined.

          • Calum says:

            Hi Ryan

            You’re right that these issues have been thrashed out here and elsewhere many times over, but I don’t think there are answers to the questions which have the status of settled truths. What’s more, most of humanity is blithely unaware.

            You’re obviously right that humans are an existence proof that consciousness – or general intelligence – is possible. But we don’t prove that AGI, i.e. artificial”, or “man-made” intelligence is possible. I certainly agree that it is probable – and likely to appear fairly soon – but doubts remain. Perhaps the level of scanning required is too granular, for instance, in the top-down approach of Whole Brain Emulation.

            As for the impact on humanity of the arrival of an AGI, You seem to think I’m saying it will necessarily be disastrous. Not so. I know that many people at MIRI, for instance, think that is the inevitable outcome unless we alter it, I don’t agree: I think we simply don’t know. There is the potential for AGI and consequent super-intelligence to be miraculous for humanity. But I also think there is a very real possibility of disaster. Which is why I think more people should be aware of what is coming.

      • iamautist
        iamautist says:

        if we can think it, we can make it real.

  • Derek Pegritz says:

    I’m really excited to see how Google’s, Facebook’s, IBM’s, and others’ individual efforts in evolving *specific* forms of digital intelligence dovetail together to form a full-scale digital sapient, kind of like a self-evolving code Voltron!

  • Carl Crenshaw says:

    follow the$$$$$$$$$$$…..AMDG

Singularity Hub Newsletter