Lenny's Podcast Video posted an update
How do AI models actually work? | Emmanuel Ameisen (Anthropic)
Why it mattersEmmanuel Ameisen explains why AI models use next-token prediction, arguing that predicting difficult text-such as a PhD thesis conclusion or a complex physics solution-can require something like understanding the world. The description contrasts this process with software built from deliberately written rules.
Discuss: How does the next-token prediction examples in the description support Ameisen's idea that AI models may develop something like an understanding of the world?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.