Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.c > #398218

Re: The secret of standard.

Show all headers | View raw

On Sun, 2026-05-03 at 02:15 +0042, yeti wrote:
> wij <wyniijj5@gmail.com> wrote:
> 
> > LLM is like a calculator, you cannot beat LLM in those elementary
> > things.  LLM's answer is generally more solid, human's is
> > 'sophisticated' or stupid.
> > 
> > Can you beat LLM in collecting facts?
> 
> Nearly every time I asked a LLM for a small code snippet to demonstrate
> a library function I wanted to understand, I got back broken code (C,
> µPy, Vlang).  The glitches were obvious and easy to fix, but I would
> expect different results from a tool "like a calculator".
> 
> E.g. once I had to tell the LLM that its answer was correct for an
> outdated Vlang version.  And a C answer hallucinated an additional
> parameter into a function call, which was there in a lot of the
> function's neighbours.  That may have been like a standard glitch for a
> Markov chain, but sure no reason for applause.
> 
> So why should I trust other facts collected by a LLM?

Baseed on the fact that Artificial Neural Network is trained by the fed data
and testing it's output (prediction or 'guess').
So, you ask questions that involve data that can be collected on internet.
Esp, provide more specific information for it to make correct matches.

You can ask LLM with your this question, it will give you fair answer.

With C problems, the answer I got from 'comp.lang.c expert' is no better than
LLM. LLM should be fast improving and I kind of not believing it is all ANN.
One thing I can assure, read-by-rote is obsolete.

Back to comp.lang.c | Previous | Next — Previous in thread | Next in thread | Find similar

Thread

The secret of standard. wij <wyniijj5@gmail.com> - 2026-05-03 08:32 +0800
  Re: The secret of standard. Keith Thompson <Keith.S.Thompson+u@gmail.com> - 2026-05-02 17:55 -0700
    Re: The secret of standard. wij <wyniijj5@gmail.com> - 2026-05-03 09:12 +0800
      Re: The secret of standard. yeti <yeti@tilde.institute> - 2026-05-03 02:15 +0042
        Re: The secret of standard. wij <wyniijj5@gmail.com> - 2026-05-03 11:55 +0800
      Re: The secret of standard. Michael S <already5chosen@yahoo.com> - 2026-05-03 08:39 +0300
        Re: The secret of standard. David Brown <david.brown@hesbynett.no> - 2026-05-03 12:28 +0200
      Re: The secret of standard. kalevi@kolttonen.fi (Kalevi Kolttonen) - 2026-05-03 19:48 +0000
        Re: The secret of standard. Michael S <already5chosen@yahoo.com> - 2026-05-03 23:15 +0300
        Re: The secret of standard. David Brown <david.brown@hesbynett.no> - 2026-05-03 22:48 +0200
          Re: The secret of standard. kalevi@kolttonen.fi (Kalevi Kolttonen) - 2026-05-03 21:45 +0000
            Re: The secret of standard. Keith Thompson <Keith.S.Thompson+u@gmail.com> - 2026-05-03 16:51 -0700
              Re: The secret of standard. Keith Thompson <Keith.S.Thompson+u@gmail.com> - 2026-05-03 17:08 -0700
                Re: The secret of standard. cross@spitfire.i.gajendra.net (Dan Cross) - 2026-05-04 00:58 +0000
              Re: The secret of standard. Janis Papanagnou <janis_papanagnou+ng@hotmail.com> - 2026-05-04 11:56 +0200
        Re: The secret of standard. Lawrence D’Oliveiro <ldo@nz.invalid> - 2026-05-05 07:19 +0000
          Re: The secret of standard. kalevi@kolttonen.fi (Kalevi Kolttonen) - 2026-05-05 08:50 +0000

csiph-web