How University Students Use Claude

I'm looking forward to the next installment on this subject from Anthropic, namely "How University Teachers Use Claude".

How many teachers are offloading their teaching duties onto LLMs? Are they reading essays and annotating them by hand? If everything is submitted electronically, why not just dump 30 or 50 papers into a LLM queue for analysis, suggested comments for improvement, etc. while the instructor gets back to the research they care about? Is this 'cheating' too?

Then there's the use of LLMs to generate problem sets, test those problem sets for accuracy, come up with interesting essay questions and so on.

I think the only real solution will be to go back to in-person instruction with handwritten problem-solving and essay-writing in class with no electronic devices allowed. This is much more demanding of both the teachers and the students, but if the goal is quality educational programs, then that's what it will take.

Alternatively, let's throw out our outmoded ideas and all get excited for an AI-based future in which professors let AI grade the essays student generate with AI.

Just think of the time everybody will save! Instead of wasting effort learning or teaching, we'll be free to spend our time doing... uh... something! Generative AI will clearly be a real 10x or even 100x multiplier! We'll spiral into cultural and intellectual oblivion so much faster than we ever thought possible!