Comments by "Stephen Villano" (@spvillano) on "The Lunduke Journal"
channel.
-
@ivok9846 there are areas in which AI excels, such as photographic analysis - especially in astronomy. Then, there's the other form of AI, Artificial Idiocy from LLM's, which confabulate a response when a search returns a result, but the result gets delayed for some reason. I don't use their preferred term hallucinate, because it is indeed confabulation of the sort one sees in dementia patients, for that very same reason - a memory recall error occurs, so the brain invents out of whole cloth a memory that seems plausible to the patient.
Worse, the AI has absolutely no clue what it is speaking about or even that it is indeed speaking at all, it's literally a Chinese Room problem by its very design, with no expert systems involved to vet its own output. There are copious examples of ChatGTP inventing case law, as one example, citing nonexistent research papers and more, rendering any output beyond untrustworthy. Yet, idiots abound who will use and trust the AI's output.
Not quite even a tempest in a teapot, more like an attempted tempest in a tepid teapot, but can't actually work up any steam.
1