Are large-language models useful? Of course they are. But it's important to really think about why, and for whom. I'm going to wade into this debate carefully because feelings are charged on both sides. (If I do this right everyone should be slightly unsatisfied with my take.) Here we go:

Thankful that we are at least starting to have more constructive debate, and this from @benjaminjriley.bsky.social is excellent When @caseynewton.bsky.social first used the phrase “fake and sucks” I could immediately tell he was referring to @edzitron.com; this piece helps clarify the args

I think this from @benjaminjriley.bsky.social is spot on, but I do wonder what would’ve happened if the student hadn’t so obviously cheated. My guess is that districts will be hesitant to discipline in cases where the evidence is less cut and dry