Dave Lane 🇳🇿 on Nostr: npub1ku76q…astrj seems a lot of people, especially in education and edtech, are ...

npub1ku76q8e9442z0jtm7e8h8p83ugx6wpuet97uxcklc9hr59txdqtqvastrj (npub1ku7…strj) seems a lot of people, especially in education and edtech, are believing the hype, and contributing to it. As I see it, LLMs (I avoid the term AI as a rule, as it's inaccurate) are simply able to make grammatically plausible interpolative regurgitations of information within a 'knowledge space', defined by a training data set. But no one seems to be curating the results & ensuring that the output is substantially (and even in nuance) accurate. Much probably isn't.