Claude’s Character

In the latest piece in his series Also True for Humans , about managing AIs like you'd manage people, Michael explores the inner workings of AI agents—the next generation of assistive AI technology—and what they need to succeed. He goes in-depth on the reason and act (ReAct) pattern, one of the first attempts to give large language models the tools... See more

"Role prompting"... telling the model to assume a role has never been a good way to elicit capabilities/style/etc. For instance, if you ask one of the Claude models to simulate Bing Sydney, assuming you can get it to consent, the simulation will probably be very inaccurate. But if you use a prompt that tricks them into predicting it indirectly (https://t.co/wJEAlPgfz6... See more