Already boosted in my timeline, but I wanted to add that I'm impressed this guy gave it far more than a fair shot, using it for over a year.
IMHO, LLMs are a parlor trick. The fact that they keep chundering up obviously wrong answers (chewing massive power doing so) makes them absolutely inappropriate for any real-world use. If I ran Google (and I should), I'd be telling the team, "Fascinating demo, but we can't possibly offer this as a paid-for service."
https://infosec.exchange/@david_chisnall/113690087142854474
=> More informations about this toot | View the thread | More toots from ewhac@mastodon.social
text/gemini
This content has been proxied by September (3851b).