Evolution, AI, and Consciousness

This is part 4 of a series. See part 1 here, part 2 here, and part 3 here.

Daniel C. Dennett is an emeritus professor at Tufts University and one of the most influential philosophers and cognitive scientists of his generation. In my podcast (see references for the full video), I recently spoke to Dennett about his recent memoir, I've Been Thinking, and his views on consciousness, free will, religion, the importance of evolution, and philosophy itself.

Walter Veit: One thinker that might have influenced you, at least outside of philosophy more than anyone else, has been Charles Darwin. So, I'm wondering, when did you first read The Origin of Species, and did it already dawn on you back then that this would change all your thinking about philosophy?

Daniel Dennett: I didn't actually read Origin of Species until after I read Dawkins' The Selfish Gene. I think it was after that that I went back and did a proper read of Darwin. And there's still, you know, some of Darwin's great books that I've only skimmed, but I realized years before that, that evolution was the key. That was one of the main themes in my dissertation in Oxford. How little I knew of evolutionary biology was embarrassing.

But I got the main idea, which is all learning is a re-design based on the design that you already have. It's a ratchet, and it's all a matter of trial and error. It's all a matter of generate-and-test, which has been the algorithm of innovation for billions of years. All we've done over the millennia, but not millions of years of science and human inquiry, is to build better, faster ways of doing generate-and-test using our thinking tools. The thinking tools that we invented using generate-and-test well, often without realizing we were doing it.

So the blind, purposeless generate-and-test of natural selection gradually turns into the deliberate, reflective, intentional generate-and-test of most of the methods in the sciences.

And now we're even learning how to re-harness natural selection itself, blind natural selection, to do further design enhancement and improvement in things like large language models and artificial life and genetic algorithms. So we're coming back full circle, and we're letting evolution do a lot of the heavy lifting for us in science these days.

Walter Veit: You've been one of the pioneers in philosophy in trying to understand artificial intelligence and accompany its development. But then we had philosophers like Hubert Dreyfus, who argued that we could never have any AI that even beats competent chess players, right? And I think that was actually a commonly held view, not only by him, but by many other philosophers.

Daniel Dennett: Yeah. Hubert Dreyfus was a very good philosopher, and we got along very well. We had some disagreements over the years including on national news television. But I always appreciated that although I thought he was wrong and making a sort of classic philosophical mistake by inflating his claims, if he had just been a little bit more modest, if he had said that AI is much harder than you think, he'd have been right for the reasons he gave. But he said it's just flat impossible, and that's wrong. And philosophers have an attraction, a fatal attraction for absolutes; something is flat impossible and inconceivable. And very often that's mistaking a failure of imagination for an insight into necessity. And that's one of the chief foibles of philosophy, I think.

So we should always sort of take a deep breath today. Instead of trying to prove it's flat impossible, let's just make sure we understand how complicated and difficult it is.

Walter Veit: Perhaps here it is worth talking about your theory of consciousness, where many philosophers, such as Thomas Nagel, insist that if we introspect into our own minds, we're presented with certainties. This is how the mind works. And perhaps this could be described as your life's work to argue against this kind of assumption.

Daniel Dennett: Yeah, exactly, and Tom has been a wonderful, loyal opposition all these years. I mean, I've known Tom since he was a graduate student at Harvard and I was an undergrad. I also gave the response to his bat paper at the Chapel Hill colloquium before it was published. So Tom Nagel and I have been sparring on these issues for decades, for more than 50 years. And I think he's a wonderful philosopher; wonderfully, usefully wrong, but thank goodness for his persistence, his inventiveness, and his clarity. I love talking with Tom about what I think he's wrong about and both in public and private.

Artificial Intelligence Essential Reads

How Leaders Can Use AI to Bring Business Objectives to Life

The Large Language Laboratory: AI as Scientist and Subject

I was very upset with his being taken in by a creationist book a number of years ago and, privately, I sort of scolded him and said, “Tom, this is a big mistake. You've had the wool pulled over your eyes.”

Walter Veit: Yeah, he did become somewhat of a hero of the creationist crowd at this time.

Daniel Dennett: That was a lapse that was all too understandable given his general mindset, but I think he really blundered on that one.