a tricky thing with integrating AI-text detection in grading or evaluation is that false positives have significantly worse impact on the writer than false negatives have on the evaluating entity

→ View original post on X — @thom_wolf,