#GenAI seems miraculous until you critically examine its output. It simply isn't trustworthy. And, because it deals in word frequencies rather than facts, rather than truth and falsehood, it's incapable of telling you when what it says is true and when it's nonsense. Explicitly telling the machine not to hallucinate, interpolate or guess will make hallucinations less frequent, but still not to the point where you can trust anything the machine says.

And, when you don't spot a hallucination, who's responsible for letting it slip through the net? Why, you are, of course.

@emilymbender @alex

Things are getting nutty. I was at my cardiologist yesterday. I have retired recently from the local hospital and we talked a little shop for a moment.
He told me the chief medical officer, the top corporate doctor in the building, was telling staff that an AI program was coming that would serve as the admitting doctor when patients came interested the hospital.

I hope people wake up before serious mistakes are made.

Thanks!

I should publish a zine called All the Ways You Can Use AI Responsibly, and then inside it just says "this page intentionally left blank".

https://dair-institute.org/maiht3k/

Will there be a transcript? (pls & thx)

I retired from my biotech job last year
but before I left, it was true that almost every single PhD, Engineer, and mid/senior manager had discovered, thru trial and error, NOT company mandated policy, that AI is a very useful and powerful tool, able to work miracles in some specific cases

not sure why my real life experience is so different from what you post seems to imply

#NoAI #ButlerianJihad

Them: But cyanide is so powerful. I’m sure there’re ways to use it responsibly.