Psychologist Paul Bloom explores whether AI can truly match human consciousness
We're living a world where AI is doing very impressive and 'very scary things,' says UofT professor
Originally published on March 3, 2023.
When it comes to favourite movies, psychologist Paul Bloom has just one: The Matrix.
"It's a wonderful movie, and it's a movie of ideas," he told The Current's Matt Galloway.
"One of the ideas is a very old idea from Descartes, probably before him, which is that we could question whether our experiences reflect a real world or whether they're illusions in some case."
While he doesn't endorse the film's idea of a simulation, the question of how the world is perceived is one that Bloom finds really intriguing.
"Although I'm not a skeptic, I think there is a world and we perceive it, and I think it evolved for a reason, but it's imperfect," he said.
Bloom, who teaches psychology at the University of Toronto, explores this and other ponderings about the mind in his new book, Psych: The Story of the Human Mind.
He spoke to Galloway about the book, as well as artificial intelligence and consciousness. Here's part of their conversation.
Do you believe in the promise of artificial intelligence? ... Do you think that we are at a point — or is there any possibility that machines would be able to replicate what we've just been talking about, the way that our minds are working?
If you asked me five years ago, I'd say no. But a lot has happened [since].
Right now, as we're talking, we're living in a world where ChatGBT and Bing are actually not only doing very impressive things, but they're really doing very scary things.
I'd say, I don't know, and I don't know a couple of things. I don't know whether these things will ever be sentient, conscious and how we'll know if they are.
I don't know whether it'll hit a wall where there'll be a certain set of things that they can't do, like, for instance, carry on a long narrative. Write not just a paragraph or a series of paragraphs, but a book.
There are people like Gary Marcus, for instance, [who] argued that the problem of these current systems is that they use all deep learning statistics, but they lack the rule-governed nature, the symbolic nature of human thought. Unless you plug that into them, they'll never be quite like a person.
We have mental representations. You have memories that you could call upon. The machines don't quite work that way, and until they work that way, I think they're going to fall short of human capacities.-Paul Bloom, psychologist
Then another thing I wonder is exactly how dangerous will they be? They could be an enormous boon to humanity, but you could imagine all sorts of very frightening outcomes.
I mean, in many ways this gets at the heart of the book. What is the difference between what those machines are able to do … and what our minds are able to do. What's the daylight between the two?
At a certain level, there is no daylight. The AI pioneer Marvin Minsky described people as machines made of meat and that's a very ugly way of talking about the people who [are] ourselves and people we love. But there's a truth to it.
But then I think when you look closer to our differences, we are different kinds of machines. These AIs, the way they're currently constructed, do statistics on the world. They're predictive machines. Basically, what they do in an extremely complicated and elaborate way is they start producing sentences and they just predict what should come next.
I think humans do more than that. I think that humans, for instance, have a model of the world. If you close your eyes, you could write down or turn your head. You could write down a piece of paper, a map of your apartment or your house, a map of the building we're in.
We have mental representations. You have memories that you could call upon. The machines don't quite work that way, and until they work that way, I think they're going to fall short of human capacities.
But we don't know … in your words, how a lump of meat in our heads gives rise to a conscious experience.
It's what's been called the hard problem compared to the easy problem of how we learn to talk and how we fall in love and all that like.
The hard problem is we know. We know that the brain is the source of the mind. We notice in a million ways, including the fact that damage to the brain affects the most intimate aspects of our thought, including the fact you could actually look at parts when they're active and then infer reliably what people are thinking.
But how a physical thing can give rise to the feeling you get when you slam your hand in a car door, when you, you know, you kiss your newborn baby or whatever … is, I think, a mystery. It's a problem of consciousness. The problem of experience.
It's driven some people to spiritual beliefs or religious beliefs or skeptical beliefs and I'm not there, but I will concede it is a problem that we are struggling with.
Is there a part of you as a guy who has studied this, that kind of doesn't want to know? Do you know what I mean? That there is a mystery in that that is tantalizing, and I like the fact that it's out there, that I can't grasp it?
If we were close to a complete theory of the mind, I might get that feeling. I might start to worry. If I could predict what other people do, well, my own world would be different. If we knew how to persuade each other with perfect scientific certainty, then I would start to worry.
But we are so far away that I think we can have great fun exploring things without running into the risk of sort of explaining everything away.
I think more generally, a lot of people see a clash between the sort of research we're discussing … and deep values that we have, like moral responsibility or or free will or spiritual value.
And I think in the end, there isn't any clash. I think these things can co-exist. We can have a science of psychology that's compatible with the values that are most important to us.
Produced by Howard Goldenthal. Q&A has been edited for length and clarity.
