AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
The Problem With Language Models
With language models, there's a baked in assumption that everything that's important is in the text. So these are not timeless pieces of truth. You know, they may seem timeless because they've been selected for a kind of timelessness. And yet they're responding to it in a way that does not make those things explicit,. Right? Like you and I use words transparently, right? If I say philosophy, then I assume that your definition of philosophy, like maps, is mainly enough on to mind that I can just say that.