"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 ... See more

tone of voice (and personality?) is the UI of the natural language / AI-agent era, and it's a pretty unexplored spectrum of possibilities at this point.