#ChatGPT-likers, especially those of you using it as a learning tool (cc @simon) how do you square the problem of it very clearly being capable of putting out "correct-looking/intelligent-sounding, but wrong" content?
I cannot see why you'd ever be able to fully trust what it's telling you is accurate, unless you're already a domain expert on the topic – at which point, why bother?
Follow
I discuss this in my latest post at https://sentientsyllabus.substack.com/p/chatgpts-achilles-heel Bottom line: it can actually be particularly useful for students if the AI is wrong in that way.
