Four robot faces indicating various emotions drawn in a 1950s comicbook style by an AI.


Happiness level


This setting will impact how snarky the LLM is in general. It's not as gradual as I'd like. AI isn't great at nuance. We know that from our poetry experiments. You can ask it about its happiness level as you interact with it and it generally replies.

Reliability level


The goal here was to try to get the chatGPT to mix fact and fiction deliberately. It has mixed results. So it's . . . unreliable.


This by far the most successful metric. The language will change although you can see the LLM struggle at times to keep words down to one syllable.

Adjust the persona inputs on the left, then submit the text generated below to start the conversation or you can click the "Generate Persona" button at the bottom of the page.