AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
How to Reconcile Your Observation With the OpenAI Narrative
I think your observation on there's not that much data to train on is counterintuitive, especially the narrative that OpenAI has created where they've spent hundreds of millions of dollars training these models. But once you find the recipe that works, like doing more of that doesn't have to be that expensive. It's kind of like you're designing a new chip or something like the design ad for takes a long time. Once you build it, you can you can stamp on more of them. And sometimes you can even sort of scale them up or hook them together to get essentially a bigger version. So on it. I think it's very interesting insight.