One straightforward way of dealing with systems like #ChatGPT or #Galactica in #higherEd is to directly confront students with an LLM-generated answer and have them critique it.
This is in fact be much better aligned with what we want to teach. Remembering and recalling facts isn’t an important skill anymore, but #criticalThinking is essential for a future in which graduates will work side by side with “#AI.” #constructiveAlignment
@tla Ah, fond memories! But I don’t think ChatGPT vs. Galactica would be that much fun; they’re both smart-asses. But running ELIZA against them might be fun: “What makes you think you are a large language model trained by OpenAI?”
@true_mxp Now I am imagining a kind of analogue to M^x psychoanalyze-pinhead, where multiple chatbots dispute in an infinite loop...