The Unpredictable New Behavior Emerging in Large Language Models

The people who build LLM's and study them don't understand a lot of what is going on. If you ask the model the right way, sometimes it'll just kind of reason out loud. Sometimes it will say, oh, it'll actually do long addition. It'll actually write out its steps. And at least for some of these tasks, GPD3 was also bad at this. That's the end of the ads. Now we're going back to the show.