…but a properly functioning calculator doesn’t confidently state the square root of -1 is pi… all current AIs make shit up when they don’t actually find a factual answer, or parrot back wrong answers from whatever crap they trained on. Until the problems of recursive ingestion decay and lack of ability to introspect on the correctness or provenance of data is solved, they cannot be trusted and they give a false sense of confidence.
the training stages? you mean every piece of science, every novel, every reference material they could possibly feed into the model? somehow is rendered useless becauuuuuse??
…but a properly functioning calculator doesn’t confidently state the square root of -1 is pi… all current AIs make shit up when they don’t actually find a factual answer, or parrot back wrong answers from whatever crap they trained on. Until the problems of recursive ingestion decay and lack of ability to introspect on the correctness or provenance of data is solved, they cannot be trusted and they give a false sense of confidence.
garbage in garbage out- if you ask good questions you get good answers, if you know how to use a calculator it can do wonderful things.
edit: 8008135
Yes
It’s way too late for that, the garbage already went in in the training stages, now you get garbage out
the training stages? you mean every piece of science, every novel, every reference material they could possibly feed into the model? somehow is rendered useless becauuuuuse??
More like Facebook posts, ad copy and, yes, b-grade fiction
oh, well… then ai will surely fade into obscurity in the coming years- let’s wait and see.
Heh