Geoffrey Hinton on AI, sentience, subjective experience, and the creativity of LLMs
Staying on Top (a free supplement to the Strategy Toolkit)
There are few people alive today who can speak about AI technology with as much eloquence, knowledge and authority as Professor Geoffrey Hinton of the University of Toronto.
And on June 25, 2025, Hinton addressed a UofT audience at the event - Frontiers of AI: Insights from a Nobel Laureate, tackling head on the existential questions of the day. Does AI exhibit sentience? Does AI have subjective experience? How creative are LLMs? He shared his position independently and then in a bit of a sparring moment with his former university intern / now founder of Cohere.ai Nick Frosst.
On sentience…
“When I get embarrassed, I go red. When AI gets embarrassed, presumably it doesn't go red. But it could still have the same behavioural aspects of embarrassment. It could still try and avoid those situations in future…”
On subjective experience…
“Suppose I've got a multimodal chatbot that has a camera.
“And it can talk, and it can point, and I train it up.
“And I put an object in front of it, and say point to the object.
“It'll point to the object, and I train it properly. Now what I do is I put a prism in front of the camera lens when it's not looking.
“And I put an object in front of it and say find the object. And then it points over there.
“And I say, no, I put a prism in front of your lens. And the chatbot says, ‘Oh, I see, the prism bent the light rays. So the object's actually straight in front of me. But I had this subjective experience it was over there.’
“The chatbot would be using subjective experience exactly like we use it.”
On the creativity of LLMs…
“For example, you have about 100 trillion connections in your brain. Gpt-4 has about a trillion. Gpt-4 knows a lot more than you do.
“So it's packed a lot more knowledge into a lot fewer connections. And to do that, it has to see relationships between all sorts of different pieces of knowledge to compress it into so few connections…
“In all the tests people have done comparing the creativity of an LLM with the creativity of a person, LLMs actually score quite well. So in a standard creativity test at, like, the 90th percentile, actually they were a couple of years ago at the 90th percentile, but probably way beyond that now.
“So I think they're actually very creative, because I think they're seeing lots of analogies between different things in order to compress so much knowledge into so few connections…
“So in terms of that creativity aspect, what does that creativity suggest about, practically speaking, what the risks are, as opposed to, like, a malicious actor? They'll be able to see creative new ways of finding people's passwords (for instance).”
Watch this video - as a strategist, you won’t regret it.