[Ozzie_osman]

But in the case of chain-of-thought and that ReAct paper, the results did have a measured increase in accuracy.

[vgel]

Oh yeah, absolutely. Just not something I'd use for, e.g., mortgages where denying someone unfairly could lead to a lawsuit.