Subj : Re: ai sucks
To   : KURT WEISKE
From : Mike Powell
Date : Sat Mar 30 2024 09:27:00

>  MP> Soon, they will be just as good at spewing out misinformation as any
>  MP> human could imagine to be.

>  I read that one of the AI scientists was concerned that LLMs (what
>  we're calling AI) rely on quality information to feed the language
>  models. Now that we're flooding the net with AI-generated content, some
>  of which is wholly inaccurate, or at the very least wildly
>  apocryphal, that information is seeding the LLMs to create more
>  innacurate AI content, and so on, and so on.

I did not even think of that... that the AI bots are also picking up stuff
that other AI bots have generated, which is making the misinformation
consumed and generated only grow.  That could have exponential results!

>  MP> It is hard enough to get a good answer out of a group of people.
>  MP> Throwing ai in there to pool a lot of information (good and bad) to
>  MP> spew out answers won't help.

>  We saw that on DOVEnet with a LLM that answered support questions.
>  Sometimes it was spot on, other times it was way off.

Now that you mention it, I do remember that.  Interactions with it could
prove quite interesting, but often not so informative.  ;)

Mike

 * SLMR 2.1a * I didn't do it nobody saw me you can't prove anything!
--- SBBSecho 3.28-Linux
 * Origin: capitolcityonline.net * Telnet/SSH:2022/HTTP (618:250/1)

.