One driver of the current flood of low-quality submissions is generative AI itself. For a problem of this scale, don’t we need a scalable response?

If we scope AI involvement appropriately, couldn’t current technology already make a meaningful contribution? Format validation, code reproducibility checks, statistical error detection, plagiarism screening—these seem distinct from “AI judging scholarly merit.”

Quantity-based evaluation certainly has its limits, but it’s also one of the few metrics that can be objectively quantified. Alternatives like research impact or societal contribution can be even more subjective. I’m curious what direction might work better here.

I appreciate Prof. Bender’s concern here. That said, Prof. Dietterich’s proposal seems to come from a practical place. Academic reviewing is unpaid volunteer work, yet submission volumes are exploding. I worry whether the system can hold up while we wait for cultural change.