Subj : Re: ChatGPT Writing
To   : Rob Mccart
From : jimmylogan
Date : Mon Nov 03 2025 20:42:10

-=> Rob Mccart wrote to JIMMYLOGAN <=-

 RM> But we are seeing many cases where it is doing things that
 RM> are intelligence oriented, like telling lies to make a point
 RM> or, as I mentioned, rather than admitting it can't find an answer
 RM> it will often give you back incorrect information.

 JI>Yeah - and I wonder if it was programmed to do that instead of admitting?
  >So as not to give an 'I don't know' answer? Because I pay the monthly
  >for ChatGPT and it will tell me it doesn't know or can't find that, etc.
  >Or it will make a mistake by telling me something, then I tell it no
  >it didn't work - and ask about something else.

 RM> My only experience with AI is that the first offers of info when doing
 RM> a search in a browser are almost always generated by an AI system
 RM> these days, and it is often a quick and correct answer if the problem
 RM> isn't too complex, but that's not a paid system where I chat with the
 RM> AI, it's just included with the browser as a search helper..

 RM> Also, if the first offered answer is questionable or obviously wrong,
 RM> I usually just carry on to offered suggestions further down the page
 RM> doing the search myself like we had to do before getting the magic AI
 RM> help rather than giving it another crack at it.

Yeah, the example you give is actually what I'm referring to. It's not
making a decision or using critical thinking skills, it's just responding
to a query based on what it has been taught to do. It is always - ALWAYS - 
regurgitating information that you could find on your own.



.... More Sugar!
--- MultiMail/Mac v0.52
 þ Synchronet þ Digital Distortion: digitaldistortionbbs.com

.