{"version":1,"type":"rich","provider_name":"Libsyn","provider_url":"https:\/\/www.libsyn.com","height":90,"width":600,"title":"Next-Token Predictor Is An AI's Job, Not Its Species","description":"I. In The Argument,  Kelsey Piper gives a good description of the ways that AIs are more than just \u201cnext-token predictors\u201d or \u201cstochastic parrots\u201d - for example, they also use fine-tuning and RLHF. But commenters, while appreciating the subtleties she introduces, object that they\u2019re still just extra layers on top of a machine that basically runs on next-token prediction. I want to approach this from a different direction. I think overemphasizing next-token prediction is a confusion of levels. On the levels where AI is a next-token predictor, you are also a next-token (technically: next-sense-datum) predictor. On the levels where you\u2019re not a next-token predictor, AI isn\u2019t one either. ","author_name":"Astral Codex Ten Podcast","author_url":"http:\/\/sscpodcast.libsyn.com\/website","html":"<iframe title=\"Libsyn Player\" style=\"border: none\" src=\"\/\/html5-player.libsyn.com\/embed\/episode\/id\/40706160\/height\/90\/theme\/custom\/thumbnail\/yes\/direction\/forward\/render-playlist\/no\/custom-color\/88AA3C\/\" height=\"90\" width=\"600\" scrolling=\"no\"  allowfullscreen webkitallowfullscreen mozallowfullscreen oallowfullscreen msallowfullscreen><\/iframe>","thumbnail_url":"https:\/\/assets.libsyn.com\/secure\/item\/40706160"}