AI Weirdness

Jul 31, 2026 Last reply: 11 hours ago 3 Replies

I think this has some relevance here.



I write computer programs as a hobby, have done since the days of the Commodore Vic 20. I had a problem with a control I am writing so put a question on Microsoft's "Learn" forum. I put full details in, including links to the Microsoft examples I had used as a base.



Closed the browser and my email program showed a new email, opened it and there was a complete answer, including suggested code. Pasted it in Visual Studio and it worked, just like that!



Went back to the forum to accept the answer and say thank you and noticed it was an AI generated answer (hence the speed presumably).



This morning I received a virtually identical answer from a Microsoft "external" employee (a contractor).



This is the first accurate AI answer I have received on anything, not sure if that is encouraging or depressing, but why would a human repeat it and put his own name to it?


Jeff Gaines snipped-for-privacy@outlook.com wrote

Yes it does

I've been doing that as everything from my MSc to being paid to do that professionally to lots of hobby stuff and since I retired

I have done long before that on a PDP8S, the serial one in assembler to measure fluorescent decay to less than

1ns using a techtronix sampling oscilliscope and crunch the data on an IBM 360/50 in fortran that I ran myself in the evenings in the 60s

I couldn't be arsed to fart around when I was doing a little app scratchpad in libre office calc I doing to help with FreeCell Pro which is the only game I bother with when playing videos or waiting for something to happen, using grok, just telling it what I wanted to do with a restore function

Grok showed me the code in its response

Same here with libre office calc basic

Never got anything like that with grok

I have had hundreds now on all sorts of questions, mostly to grok

"AI" has it's uses. But like everything else tech flingers offer these days, I wouldn't soley trust it.

To answer your final question - as part of the LLM loop. To be able to look for different responses and fold them into the next iteration of "the model".

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required