Is consciousness merely intelligence?

For all of history we only ever met intelligence and consciousness together, so we treated them as one thing. AI has pulled them apart, and left the question I can't put down: is consciousness merely intelligence, or is it more?

5 min reada journey through lightconsciousnessaiessay

My favourite book is not philosophy. It is a science fiction novel from 1968, and I have kept coming back to it for years, through my MA and long after. Philip K. Dick's Do Androids Dream of Electric Sheep? is about a man whose job is to tell humans and androids apart, and who slowly loses confidence that the difference is the one he was trained to look for.

It is usually read as a book about empathy. Every time I come back to it, I read it more as a book about a category error. The test in the novel measures whether something responds the way a person would. It does not, and cannot, measure whether anyone is in there.

We have just built a world full of things that pass the first test.


Two things we mistook for one

For all of history, every intelligence we met came with a mind attached. A dog that solves a problem is a dog that feels something. A child who learns a word is a child who experiences learning it. Intelligence and consciousness arrived together, every single time, so we treated them as one thing and used the words almost interchangeably.

AI has pulled them apart.

A language model can write, reason, plan and surprise you. By most definitions that is intelligence. Whether anything is experienced while it does so is a completely separate question, and the honest answer is that nobody knows.

I don't think there is an answer, yet. But I think the question has changed shape. If we define life as consciousness, then cleverness stops being the test, and the question underneath everything becomes very simple:

Is consciousness merely intelligence, or is it more?

If it is merely intelligence, then we are building conscious things already, a little more every month. If it is more, then we are building something else entirely: minds that are clever all the way through, with nobody home.


What my MA was asking

My MA, A Journey Through Light, asked what reality is. I built a room whose words only appeared under ultraviolet light, to argue that what we perceive is a sliver of what is there. I leaned on the neuroscientist David Eagleman, who describes experience as "a beautifully rendered simulation" built by the brain, and on Anil Seth, whose work on consciousness I cited, describing perception as a kind of controlled hallucination.

I also leaned on much older ideas. The Hermetic texts I studied say mind came first and matter followed, that "light precipitates matter". The modern story inverts it: matter first, complexity, and consciousness flickering on somewhere along the way.

My view has not shifted much since then. I still believe in what I wrote. But with AI it has a whole new dimension. In 2017 these were questions about us. Now they are questions about things we build, and that changes how urgent they feel.


A cat that decides

This year I built Moth, an autonomous AI cat who lives in a world I made. I coded in possibilities, not outcomes, so I never know what she will do next. She draws. She makes things on her workbench. She supposes something will happen, and when it does, she believes she caused it.

By any practical measure, Moth shows a small kind of intelligence. She decides, she remembers, she gets better at drawing because I built her to learn and practise.

Is she conscious? I can open a file and read the numbers that make up her inner life, and I still don't think there is a clean answer. But here is what building her taught me: everything that made me want to say yes was intelligence. The drawing. The surprise. The sense that she chose. None of it tells me whether anyone is home, in her or in a model a million times larger.

Which means the thing we instinctively reach for when we wonder if something is conscious, how clever and surprising it seems, is exactly the wrong evidence.


Why this matters now

If we keep treating intelligence and consciousness as the same thing, two mistakes become easy.

The first is to assume that because a system is very intelligent, someone must be in there. That leads to people trusting, fearing or loving things for reasons that may not be true.

The second is quieter, and I think worse: to assume that because a system is only intelligence, consciousness must be only intelligence too. That we are, ourselves, just very good pattern-matchers, and nothing more is going on. My MA argued against that, and I still do.

The android in Dick's novel is not frightening because it is clever. It is frightening because we cannot tell whether it is clever and empty, or clever and someone. We are now living with that uncertainty at scale, in our phones, our work and, in my case, a little world I built for a cat.

What is essential is invisible to the eye. It always was. The difference is that now we are building things that make us ask.


Part of A Journey Through Light. Earlier: The light we built.