You of course have your opinion.
But the fact is, before yesterday, I designed a perfectly valid
machine—it will be cut and assembled. These new technologies are just
tools, like everything else.
You cannot generalize and reject it generally. You can try, but you
can't. Millions of people have recognized values of those tools. It is
now irreversible.
You can hand a welding rod to someone and watch them make a mess, but a
real welder is an artist: with that same rod, they can build almost
anything. The tool doesn't make the work; the skill, intent, and
understanding do. New generative tools are no different—it can produce
slop or insight, depending on who wields it and how carefully they check
the result.
Many people, including me, cannot express themselves accurately,
conclusively, specifically, the skill of expressing oneself is valuable.
Generative LLMs have the skill of expressing well the generated text.
It implies, if you are like me, you may lack the skill to express
yourself with enough context, specific, accurate, conclusive, and then
you may not get the right output.
Another thing is just as learning, if you do not know why would you
learn and how to apply it in life, then why would you learn at all?!
Same with generative LLMs, why would you use it if you do not have
practical application for it in life? So your specific and individual
case should not become standard for all the other people. "You see
people, I don't use AI". So nobody should use.
Welding rod is not used by many people. Does it mean nobody should use
it? Yet you are living in the house that required some welding, or
driving a car that required welding.
There is immense use of generative LLMs. Truly they are not at level to
make magical things by reading your mind, but many of them, including
locally run models can provide and pin point problems and help human
faster to isolate issues, and to resolve issues.
That is undeniable fact as of June 2026.
--
Jean Louis
On 2026-06-19 05:19, NIIBE Yutaka via libreplanet-discuss wrote:
Hello,
Let me share some of my thoughts. I think that the points Mr. Jean
Louis suggested (in English) are relevant.
Akira Urushibata wrote:
The recent message by Mr. Jean Louis contains much Japanese text.
It is AI-assisted/generated translation, I suppose. And it would
suggest we need to be careful for the technology.
(IMHO, use of AI tool should be explicitly addressed.)
If you speak Japanese, you can easily find it's strange text. For
others, it would look possibly "great". And even if it looks strange,
occasionally, it has good points.
We have similar situation in security reports for software. These
days,
it's common that Free Software Projects receive AI slop reports. For
experts, it's just annoying, but we can't ignore because: even if it
looks strange, occasionally, it has valid points. (In most cases,
reading such reports is a waste of time.)
Well, as one of Free Software developers, indeed, we have some
problems.
I list some and put short comments of mine.
* AI scrapers
Our development site, dev.gnupg.org, suffers from AI scrapers, a lot.
I
think that it's not only our site.
* AI slop reports
I think that experts themselves should proof read AI-assisted/generated
reports carefully and make sure the point claimed is valid before
sending out.
* AI-assisted/generated code and use of proprietary services
In GCC mailing list, Alexandre Oliva wrote a reply and described it
"poison". I'd admire his counter action.
P.S. I didn't use any AI for the text above. I use GNU Emacs for
editing text and GNU Aspell for spell checking.
_______________________________________________
libreplanet-discuss mailing list
[email protected]
https://lists.libreplanet.org/mailman/listinfo/libreplanet-discuss
