Re: Discussion: Using LLMs the Right Way: 10/1/2025 7pm Eastern Daylight time

Don Marti <[email protected]> Fri, 3 Oct 2025 16:56:26 -0700
Newsgroups gmane.org.user-groups.linux.cabal
Message-ID <[email protected]>
This is a multi-part message in MIME format.
--===============0679433111==
Content-Type: multipart/alternative;
 boundary="------------bShHOblF0uk0in056OdgXuRG"
Content-Language: en-US

This is a multi-part message in MIME format.
--------------bShHOblF0uk0in056OdgXuRG
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: quoted-printable

On 10/3/25 13:36, Deirdre Saoirse Moen wrote:

> LLMs are a parlor game that use far more resources than they can provid=
e value to society. They don=E2=80=99t =E2=80=9Cwork=E2=80=9D in the sens=
e that they are not intelligent, and are not true AI. At *best* you will =
get a junior dev out of them, but a junior dev who makes mistakes like de=
leting code that shouldn=E2=80=99t be deleted. Many people do not have en=
ough exposure to LLMs to realize they are basically attempting to recreat=
e that kid who constantly screws up then pretends nothing bad happened, t=
hen wonders why you stopped talking to them.

I was talking about a web publishing/advertising thing that's fairly=20
complicated and someone said they would use ChatGPT to get some=20
background explanation of it.=C2=A0 That seemed like a bad place to start=
,=20
for two reasons.

1. It's about a market in which a lot of people have an incentive to=20
persuade others that things are a certain way, so there's a lot of sales=20
copy from one point of view.

2. The technology is changing fairly quickly, so most of the content out=20
there would apply to older code or practices.

> That said, I did find it useful to be able to interactively ask questio=
ns that were specific to $FIRM=E2=80=99s processes and have a live answer=
 that was usually at least close enough to correct I could find the answe=
r from where it led me.
Now I'm wondering about the categories of things for which an LLM would=20
be useful.=C2=A0 Syntax for doing something that appears in publicly=20
available scripts might be a good example -- I have been able to "vibe=20
code" some correct Python code to extract material from an HTML file=20
with Beautiful Soup, and get the jq(1) syntax right for getting a subset=20
of the info from a big JSON file. (For both of those I could test the=20
output, though)
--------------bShHOblF0uk0in056OdgXuRG
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE html>
<html>
  <head>
    <meta http-equiv=3D"Content-Type" content=3D"text/html; charset=3DUTF=
-8">
  </head>
  <body>
    <p>On 10/3/25 13:36, Deirdre Saoirse Moen wrote:</p>
    <blockquote type=3D"cite"
      cite=3D"mid:[email protected]">
      <pre wrap=3D"" class=3D"moz-quote-pre">LLMs are a parlor game that =
use far more resources than they can provide value to society. They don=E2=
=80=99t =E2=80=9Cwork=E2=80=9D in the sense that they are not intelligent=
, and are not true AI. At *best* you will get a junior dev out of them, b=
ut a junior dev who makes mistakes like deleting code that shouldn=E2=80=99=
t be deleted. Many people do not have enough exposure to LLMs to realize =
they are basically attempting to recreate that kid who constantly screws =
up then pretends nothing bad happened, then wonders why you stopped talki=
ng to them.</pre>
    </blockquote>
    <p>I was talking about a web publishing/advertising thing that's
      fairly complicated and someone said they would use ChatGPT to get
      some background explanation of it.=C2=A0 That seemed like a bad pla=
ce
      to start, for two reasons.</p>
    <p>1. It's about a market in which a lot of people have an incentive
      to persuade others that things are a certain way, so there's a lot
      of sales copy from one point of view.</p>
    <p>2. The technology is changing fairly quickly, so most of the
      content out there would apply to older code or practices.</p>
    <p><span style=3D"white-space: pre-wrap">
</span><span style=3D"white-space: pre-wrap">
</span></p>
    <blockquote type=3D"cite"
      cite=3D"mid:[email protected]">
      <pre wrap=3D"" class=3D"moz-quote-pre">That said, I did find it use=
ful to be able to interactively ask questions that were specific to $FIRM=
=E2=80=99s processes and have a live answer that was usually at least clo=
se enough to correct I could find the answer from where it led me.</pre>
    </blockquote>
    Now I'm wondering about the categories of things for which an LLM
    would be useful.=C2=A0 Syntax for doing something that appears in
    publicly available scripts might be a good example -- I have been
    able to "vibe code" some correct Python code to extract material
    from an HTML file with Beautiful Soup, and get the jq(1) syntax
    right for getting a subset of the info from a big JSON file. (For
    both of those I could test the output, though)
  </body>
</html>

--------------bShHOblF0uk0in056OdgXuRG--


--===============0679433111==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
conspire mailing list
[email protected]
http://linuxmafia.com/mailman/listinfo/conspire

--===============0679433111==--