@realn2s I said elsewhere that I think the main goal of the paper is to reinforce the notion that the speed and volume (regardless of quality or correctness) of production is a necessary factor to consider related to AI, and that more is necessarily and unambiguously better.

So, that it seems to interfere with learning must be a "tradeoff" and not a "deal breaker". To the extent this study can support any conclusion at all, that is. I think it's a quite a bad study.

Additionally, it seems like a move toward blaming people for their own deskilling. If there's a way to use these things and _not_ suffer learning loss, then anyone who _does_ must be individually to blame.