Sycophancy in AI

Thanks to Thomas for working hard to grind the sycophancy out of the toolbox tools.

If you ever wondered if the base models are giving you exaggeratedly positive feedback, check this out: