[scottporad221]

This is cool because it kind of gets at the core of transparency about AI models.

I would imagine if this would built a certain way it could not only tell a brand what I models are saying, but could be used to look for bias in models that do credit scores, for example.

P.S. I also used to work at Rover!

[batterylake]

I agree, it kind of reminds me of this paper that shows LLMs, just like humans, will preferentially remember information, which can lead to biased outputs.

https://www.pnas.org/doi/abs/10.1073/pnas.2313790120