Douglas Hofstadter

Neither Douglas Hofstadter nor Daniel Dennett are easy writers to read quickly. Dennett’s Darwin’s Dangerous Idea and Hofstadter’s Surfaces and Essences are two of the most demanding books I’ve picked up in the recent past. Luckily, in The Mind’s I, an effort that combines both their talents, they find a way to better let their readers in. This book looks at the philosophical concept of the self—how a mind views itself—through the writings of other people. Hofstadter and show more Dennett use historic and imaginative accounts written by Jorge Luis Borges, Alan Turing, Richard Dawkins, and many others as points of reflection from which they can get into their intended philosophical discussions. This helps accomplish two very interesting goals: pointing the reader towards other authors they might not have known before and helping the reader through some of the more complex thought experiments surrounding the concept of the self. All throughout the book there are smatterings of philosophy, fiction, physics, and even free will. They manage to steer clear of the more tautological loops that philosophy sometimes falls in to, and in the end, arrange a very good book that makes the reader think deeply without straining themselves. An intense but intriguing read. show less

Can human thought processes be emulated artificially? This is very much a "how" question and Hofstadter explores it as such, looking for the most promising way forward. He believes it lies in the study of self-reference and tackles this from multiple angles across a variety of disciplines (music, art, mathematics, computer programming, molecular biology, etc.), on the theory that without self-reference you can't arrive at self-awareness and full consciousness or actual thinking. He makes the show more discussion a fascinating one in how ably he ties things together that I wouldn't have thought had much in common.

Jumping to the end, much of the concluding chapters demonstrate the extremely complicated and multi-layered processes that emerge when you break down even the most simple mental processes in humans, demonstrating how very difficult it is to emulate these processes artificially. Hofstadter seems wedded to the idea that artificial intelligence means closely emulating human thought and how it is arrived at. Getting to the same (or improved) results by any different means doesn't seem to count as true AI in his opinion. He tries to prognosticate what artificial intelligence will and will not be capable of in future and you could say that he still isn't wrong, pointing to his own definition. The AI that he envisions is not GenAI like ChatGPT but something that still lies well beyond us, a computer that really does think (and feel?) the same independent way that a person does, and consequently has many of our same flaws as well. Whether computer science today or in future sees any utility in trying to realize the AI of Hofstadter's vision is doubtful. To me it seems fair to weigh his 1979 vision against the direction things are actually going in 2026 and say that he has already missed the mark in several respects, all stemming from his missing the "why" that industrial science has actually been motivated by.

What we are popularly defining as AI today is function-oriented, and possibly all it will ever be. While it can be increasingly made to look and sound human, it will never "think" because that isn't its objective. This book remains brilliant for its wonderfully entertaining way of exploring the possibility of something beyond that - like Commander Data from Star Trek - which perhaps some niche areas of AI study are still pursuing, and how they might get there. show less

This is certainly the greatest book of popular science I've encountered or heard of yet--it's accessible, engaging, playful, but also very deep and original in its analysis. It gives brief intellectual history where necessary, and repeats the same arguments in many surprisingly different ways for comprehension.

But its ultimate ignorance/dismissal of the social world, assertion of objective meaning, and computational theory of mind force me to give it a lower rating.

I have been reading this book, a bite at a time, for more than four years. I'm quite nearly through it now, and I think I may mourn when I read the last pages.

I can wholeheartedly recommend this if you are a fan of Dan Dennett's work, of Zen koans, or of other writing which addresses questions wrongly asked. GEB masterfully breaks down common (or at least, my) preconceptions about abstraction and consciousness.