“Connor Leahy on Aliens, Ethics, Economics, Memetics, and Education § GPT-4”, Connor Leahy2023-01-19 (GPT-4, Codex)⁠:

Timestamps:

• 00:00 Introduction

• 01:00 Defining artificial general intelligence

• 04:52 What makes humans more powerful than chimps?

• 17:23 Would AIs have to be social to be intelligent?

• 20:29 Importing humanity’s memes into AIs

• 23:07 How do we measure progress in AI?

• 42:39 Gut feelings about AI progress

• 47:29 Connor’s predictions about AGI

• 52:44 Is predicting AGI soon betting against the market?

• 57:43 How accurate are prediction markets about AGI?

…Q: Um so what will be the most impressive thing that GPT-4 will be able to do?

Connor Leahy: Oh no I’m using up all my trading alpha! [referencing earlier discussion about whether one could trade profitably on knowledge of DL advances] No, um, I don’t know what the most impressive thing would be, because I expect once people have it in their hands they’re gonna like, you know, elicit more and more impressive things. I expect Riley Goodside is going to make it do some pretty amazing things pretty quickly… um…

One of the things I know it can do is that it’s going to be able to write much more sophisticated software than like the current systems can do. Like, ChatGPT is great for like snippets and stuff, but it can’t write whole programs before.

Q: That’s, that is actually, that could change things substantially, yep.