What’s better than artificial intelligence? Why, Personal Superintelligence, of course. Meaning, technology that’s more intelligent than a human, but also serving humans. That’s what Meta (the company behind Facebook, Instagram, and WhatsApp) has announced it’s working on.

Mark Zuckerberg is serious about Meta Superintelligence Labs (MSL). In recent months, he has hired some of the biggest AI guns, poaching top talent from most of his rival AI giants with huge salary offers. He also has the advantage of vast troves of data that Meta already has on individuals, collected whenever we interact with their services, or even with apparently unrelated online activities. The files of data it holds on most of the world’s population — even those without Facebook accounts — underlie the personalised advertising which is Meta’s main source of income. And this is key to the “personal” bit of Personal Superintelligence.

Mark Zuckerberg’s vision is of “everyone having a personal superintelligence that helps you achieve your goals, create what you want to see in the world, experience any adventure, be a better friend to those you care about, and grow to become the person you aspire to be”.

This sounds suspiciously like the kind of AI chatbot assistant that Meta — along with other technology companies — is already rolling out on your laptop or smartphone.  But it is also a canny business move, distinguishing Meta’s project from centralised AI models. Zuckerberg’s announcement letter invokes the dystopian vision of a world where AI has taken everyone’s jobs, and we “live on a dole of its output,” in his words. Instead, he promises, each of us will have a personal superintelligence “that knows us deeply, understands our goals, and can help us achieve them”.

Whether a machine can ever know a person, or understand human goals, is very doubtful, but Meta’s AI will have a great deal of data to help. Interacting with an AI personal assistant will not only add to the in-depth information Meta holds, but also create a feeling of trust, potentially laying the user open to influence on what to buy, how to behave, or even what to believe.

The goal of AI that can out-think humans has haunted and driven technologists for many years. Alongside the heady promise of a human-created being that outstrips its creators run the fears of such a powerful technology escaping human control. Given our limited understanding of our own minds, it is hubristic to imagine we can create Artificial General Intelligence — a machine that matches and surpasses human thinking — but in many specific areas AI already out-performs humans. You don’t need to believe in a synthetic mind superior to our own, to see the potential power of AI.

One of Meta’s new hires to lead the Superintelligence Lab, Alexandr Wang, with ex-Google CEO Eric Schmidt and Dan Hendrycks of the Centre for AI Safety, recently published a paper comparing superintelligence to nuclear weapons. Governments already recognise its importance, strategically as well as economically. The race for ever-faster, ever-smarter AI is global.

To this competition, Meta brings intellectual firepower, the physical infrastructure of computing power, and the resource of massive banks of data on which to train AI models. It could soon be launching its Personal Superintelligence in a device near you. But in spite of the massive investment in people, computing power, and data processing, it’s still not clear what makes it qualitatively different from existing AI assistants. Faster, less prone to hallucinations and stupid mistakes, and equipped with astonishingly intimate details of your life, yes. Smarter than a human? No.

Mark Zuckerberg calls it “a tool for personal empowerment”, an AI that will help you instead of taking your place. But, like so many of Meta’s activities, it’s not clear whether we, the users, are intended to be the customers, or merely the audience for an advertising platform more personalised, and more all-pervading, than ever before. If you thought Facebook knew too much about your habits, friends, and innermost thoughts before, just wait till your entire digital life is being lived through a personal AI assistant, “superintelligent” or not.

---

Timandra Harkness presents the BBC Radio 4 series, FutureProofing and How To Disagree. Her book, Technology is Not the Problem, is published by Harper Collins.

TimandraHarknes

---