A Revolution in Cognition?

It is said that we live in the period of the Fourth Industrial Revolution, a period where there is fusion of technologies across domains — digital, physical, and biological — and where the boundary between tool and agent starts dissolving. I think we are seeing the start of a revolution that goes beyond the tools and agents, and that shifts towards the driver: cognition.

The industrial revolutions have been, at heart, a story of extension — and extension is not a human invention but life’s oldest habit: living things extend their capabilities through artefacts, non-living and living alike — shells and webs and dams, domesticated species, each other. Sara Walker makes the point at full depth: life is the process that constructs such objects, and our technology is not outside that process but its continuation. Seen that way, mechanisation extended us past muscle — the loom, the engine, the assembly line. Electrification extended the reach — power no longer needed to stay local. The information cycle extended our handling of data — storing it, moving it, processing it at scale. And alongside the physical runs a much older mental thread: our brain is an ingenious unit, but severely resource-constrained, and over the centuries we have extended its reach through artefact after artefact — writing, printing, recording, calculators, computers.

The latest wave brings something different in kind. Neural networks give us the ability to process information the way no rule-book ever could, and with that comes the ability to hand our artefacts not calculation but substantial amounts of thinking — the lighter, but often very time-consuming, forms of mental activity. This is a paradigm shift.

I mean that in a precise sense. The shift is, in my opinion, of the same magnitude as the move from classical physics into quantum mechanics and relativity — and comparable in nature, not just in size. Then, a physics built on a determinate, observer-independent picture met phenomena that refused to fit it, and what counted as understanding had to change. Now, centuries of science and philosophy built on rational, logical thinking — explicit rules, transparent chains of reasoning — meet the fact that the machines that finally work at thinking are not logic engines at all. They are associative. The paradigm shift forces our attention towards an associative way of processing information, and away from the assumption that the rational-logical way is what thinking is.

That’s self-evident, I hear you say — how else? But there is a difference from the earlier waves. Every earlier extension gave us something we did not have; this time the artefact is a slice of what we ourselves are. Having this associative thinking trained on such vast output of human thinking makes it very powerful on one thing — so powerful that one can easily forget it is still far short of the brain on everything else, and the brain does all of that on a tiny budget.

And that is where the misreading comes from. We see the amplified capacity and read the whole brain architecture into it. More-than-us on one axis gets mistaken for a match for us on all of them. Hence the chatter about AGI, superintelligence and AI taking over the world. Don’t be fooled — the danger lies in humans deploying it inappropriately, letting AI loose without an appropriate support structure.

Should a company hire a new trainee and give them full independent authority on their first day? The trainee may be very clever but still lacks experience. They need the support of a team.

The power of AI calls for humility on our part. We approached rational thinking as the superior form of thinking and left cognition in the broad sense — associative included — aside. The attention went elsewhere: to the mind as the experienced end-product, and to the brain as the organ. AI’s thinking is associative, so we had better understand it.

The revolution worth having is to understand cognition from the ground up, not only in AI but equally in life in general.

This post is part of the conversations series.

Photo: stbuccia / Unsplash