Hello = “Hello”
world = ”world”
print(f”{Hello} {world}!”)

And I especially applaud your decision to call for ending the use of the thought terminating cliche “AI slop”.

Great piece! Instead of papers, contributions should instead be added into a larger data structure that can be more easily traversed by a human or AI. Think better clustering of ideas, results etc. It was always inefficient to image humans keeping up with the volume of publications…

Sorry, but your triumphant claims seem exaggerated. If self driving cars were causing so many accidents the media would be crowing around it from the roof tops me thinks… because AI is bead sells

And the teacher‘s role would absolutely still be crucial to ensure the interactions are meaningful. I think sticking our head in the sand in school and pretending they don’t exist would be a disservice to the students as well.

Im not sure that’s strictly true. I agree that the systems available now are likely too generalized to be good out of the box in a school setting. But I’ve been successful in getting them to assess my own though process. I image specialized didactic versions could be created.

my counter is that given the age and pace of change of the technology it seems premature to produce definitive research into it‘s impacts. Thank you for the conversation. I’ll go back to watching for now.

Good points. I can’t currently justify the energy expenditures either, not while the fuel is dirty. But i see no problem using renewable energy once it becomes the dominant source. the water use could be improved with recycling.

Of course, Im not pretending it will would work out of the box and there is no real replacement to writing an essay. But it could certainly asses an essay written by a student and offer feedback and ask follow up questions.

The LLM can ask questions back. Now it already does something like this when it tries to steer the discussion. Im thinking offshoots of the current agents could be tailored to be more didactic.
The issue with much of the criticism of LLM is outright dismissal instead of curiosity.

Of course, that’s why the systems would need to be pre prompted to avoid sycophancy. the teacher can then asses the quality of the exchange. I’m not implying it’s a perfect application but just an example where there could be promise.

I mean the ”Socratic method” is all about learning by asking questions. LLMs seem like a good interlocutor to me with proper pre-prompting by the teacher.

Do you have a method for reliably distinguishing between “LLM mistakes“ and plain “author mistakes“? the reviewers job is to assess the quality of the work based on whether it works and is valid. Not who or what produced it…?

They’re using it wrong is my best bet.