Dwarkesh Patel: recommendation

Transcript context

…No. The people who are working on AI right now are huge fans of yours. They're the ones who initially recommended the book to me because the way they see the progress in the field reminded them of this book. Because you start off with these initial scientific hints. With deep learning, for example, here's something that can teach itself any function is similar to Szilárd noticing the nuclear chain reaction. In AI there's these scaling laws that say that if you make the model this much bigger, it gets much better at reasoning, at predicting text, and so on. And then you can extrapolate this curve. And you can see we get two more orders of magnitude, and we get to something that looks like human level intelligence. Anyway, a lot of the people who are working in AI have become huge fans of your book because of this reason. They see a lot of analogies in the next few years. They must be at page 400 in their minds of where the Manhattan Project was. We must later on talk about unintended consequences. I find the subject absolutely fascinating. I think my next book might be called Unintended Consequences.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.