Mollick: AI has no constraints, so the way research is done needs rethinking

Eva Vivalt has released a new working paper exploring whether AI can predict treatment effects in economics based on pre-registration documents, and the conclusion is "yes." Ethan Mollick commented on this: scholars invest enormous effort in writing a paper, so they often choose topics expected to produce interesting and significant results, which also pushes research in a predictable direction; AI has no such constraint, forcing people to rethink many practices in scientific research.