| Sysop: | Amessyroom |
|---|---|
| Location: | Fayetteville, NC |
| Users: | 27 |
| Nodes: | 6 (0 / 6) |
| Uptime: | 40:39:22 |
| Calls: | 631 |
| Calls today: | 2 |
| Files: | 1,187 |
| D/L today: |
24 files (29,813K bytes) |
| Messages: | 174,719 |
When I ask ChatGPT a question about FreeBSD, I very often get the
wrong answer. Does FreeBSD have a plan for improving the quality of
answers so that it can be taken into account in the next version?
Hello,-aI'll bite, even though it's not the appropriate mailing list.
When I ask ChatGPT a question about FreeBSD, I very often get the wrong answer. Does FreeBSD have a plan for improving the quality of answers
so that it can be taken into account in the next version?
On Oct 11, 2025, at 4:14rC>AM, ft <ml@ft-c.de> wrote:
Hello,
When I ask ChatGPT a question about FreeBSD, I very often get the wrong answer. Does FreeBSD have a plan for improving the quality of answers
so that it can be taken into account in the next version?
Here are some suggestions (which I received from ChatGPT) on how this
could be done:
First, use the built-in feedback system. If you are using ChatGPT in a browser, you will usually see thumbs-up and thumbs-down buttons under
each answer. If an answer is incorrect, click the thumbs-down button.
In the feedback form, explain what is wrong and provide the correct information. For example, if ChatGPT suggests an invalid option for the
pkg command, you could write: rCLThe response about pkg usage is
incorrect. The option -b does not exist as described. The correct
option to build a package is -p, as documented in man pkg(8).rCY
Second, it would be great if the FreeBSD community could create a list
to collect and document wrong answers. Keeping a simple personal record
of mistakes is also very helpful. The idea is to systematically
document any wrong answers before submitting them as feedback or
sharing them with others. This makes it easier to prove what was
incorrect, ensures you donrCOt forget important corrections, and helps
others understand the issue clearly.
A practical way to do this is to note:
- The topic, command, or function the answer relates to, e.g., rCLpkg buildrCY
- The incorrect answer given by ChatGPT, e.g., rCLOption -b builds a packagerCY
- The correct answer, e.g., rCLOption -p builds a packagerCY> - The source or reference, e.g., rCLman pkg(8)rCY
- You can collect this information in a simple text document or table.
Later, when giving feedback to OpenAI, you have all the information
ready and verifiable.
Third, use community channels to amplify the feedback. OpenAIrCOs
community forum (community.openai.com) allows you to create threads
about repeated errors or FreeBSD-specific issues. Other users can
comment, verify, or add additional information. For open-source AI
models, you can submit issues directly on GitHub or the projectrCOs issue tracker.
Finally, when giving feedback about FreeBSD, try to provide context.
Include the exact FreeBSD version, quote commands or man pages, and
explain why the original answer is incorrect. Logs or screenshots can
also be helpful.
By following these steps, we can gradually improve the quality of AI responses for FreeBSD topics and help future users get more accurate information.
Best regards,
Franz
On Sat Oct 11, 2025 at 1:14 PM CEST, ft wrote:Agreed. A very old music hall joke :
Hello,-a
When I ask ChatGPT a question about FreeBSD, I very often get the wrong answer. Does FreeBSD have a plan for improving the quality of answers
so that it can be taken into account in the next version?
I'll bite, even though it's not the appropriate mailing list.
Why should we take any LLM into account for the next version? We have documentation (man pages, handbook, other docs), and real people you can speak to. There is no reason to waste already scarce resources to
improve ChatGPT.
Christos
Sat, 11 Oct 2025 14:55:11 +0200 "Christos Margiolis" <christos@freebsd.or=ng
wrote:
"Christos Margiolis" wrote:
On Sat Oct 11, 2025 at 1:14 PM CEST, ft wrote:
Hello,=C3=82
When I ask ChatGPT a question about FreeBSD, I very often get the wro=
nanswer. Does FreeBSD have a plan for improving the quality of answers
so that it can be taken into account in the next version?
I'll bite, even though it's not the appropriate mailing list.
Why should we take any LLM into account for the next version? We have documentation (man pages, handbook, other docs), and real people you ca=
speak to. There is no reason to waste already scarce resources to
improve ChatGPT.
Christos
Agreed. A very old music hall joke :
"Doctor! Doctor! It hurts when I do this ... ! "Well, don't do that
then!"
ie Don't expect quality AI about small moving targets (eg FreeBSD).
AI will do better against static large targets, where a vast general
public can be tempted to contribute responses free for AI to hone improvements.
That's not us.
AI is proving useful to smaller groups eg doctors, ("Did you also
consider the rare XYZ disease?") I assume theres lots more doctors
in the world than FreeBSD users, & some doctors will see it as part
of their paid job to correct false responses from AI.
That's not us.
Unpaid volunteer time seems better spent improving FreeBSD & docs, not inefficiently correcting old AI mis-understanding of a moving target.
If/ when AI learns source code & SVN etc, all bets off though - Shudder !
Cheers,
--
Julian Stacey http://berklix.org/jhs/mail/ @gmail blocks replies.
Arm Ukraine. Contraception V. global warming. Israel starves Gaza.
When I ask ChatGPT a question about FreeBSD, I very often get the wrong >answer.--
When I ask ChatGPT a question [...] I very often get the wrong answer.Don't do that, then.
ft <ml@ft-c.de> writes:|
When I ask ChatGPT a question [...] I very often get the wrong answer.
Don't do that, then.
Artificial intelligence exists only in science fiction. ChatGPT is not artificial intelligence, it is a deterministic computer program that
uses a large statistical model of natural language to complete text submitted to it. Since it has been trained on a corpus that includes
many examples of questions followed by answers, when prompted with a question, it will produce something that has the shape of an answer, but absolutely no effort has gone into ensuring that the answer is correct,
nor does anyone involved have any idea how to even begin doing that.
wrote:r.
Michael Butler wrote in
<d716c704-3ae7-4c6a-a37b-fcecc3b743cf@protected-networks.net>:
|On 10/13/25 10:23, Dag-Erling Sm=C3=B8rgrav wrote:
ft <ml@ft-c.de> writes:
When I ask ChatGPT a question [...] I very often get the wrong answe=
otDon't do that, then.
Artificial intelligence exists only in science fiction. ChatGPT is n=
t,artificial intelligence, it is a deterministic computer program thatbut
uses a large statistical model of natural language to complete text submitted to it. Since it has been trained on a corpus that includes many examples of questions followed by answers, when prompted with a question, it will produce something that has the shape of an answer,
absolutely no effort has gone into ensuring that the answer is correc=
e\nor does anyone involved have any idea how to even begin doing that.|
|Spotted this gem on another mailing list recently ..
|
|All that *ANY* LLM can provide here is a statistically
|less-improbable-than-random stream of words that may or may not include
|any statements of true facts, uninfluenced by anything resembling knowl=
|dge.
Now i have to say one thing. I never tried AI, i do not like AI,
because it is used and trained by the same old dumb humans, not
to talk about Kubrick's HAL, and of course Asimov, i *think* AI
should be a scientific thing that is carefully developed before
it enters "the normal world", maybe even so that dedicated wind
and solar farms are built in order to drive the AI then used in
"the normal world".
Having said that. I recently opened a ChatGPT instance to read
the conversation initiated by the Field medalist and otherwise
Hyper Mathematic whose name i have forgotten (not a mathematician
here) who tried it out in order to address a problem asked by
someone on some "stackoverflow-alike-thing for mathematicians".
It was about proof that, iirc, "the sum of dividers of a number
is always larger than the number itself".
Now i tell you, that shitty conversion was fascinating, not that
i understood mostly a single mathematical term they were throwing
back and forth, and i would not post this message if then, and
here i was stunned and still, one answer that cames back was like
about "love this topic" or similar. And then it crushed to super
detail, and in the end it presented a small python reproducer.
In the hands of, and in correspondence with this math genius the
AI worked in an amazing way, with turns that have shown
a thrilling -- as i with my own restricted capabilities think --
topic reflection. This does not counteract the first paragraph.
--steffen
|
|Der Kragenbaer, The moon bear,
|der holt sich munter he cheerfully and one by one
|einen nach dem anderen runter wa.ks himself off
|(By Robert Gernhardt)
Michael Butler wrote in
<d716c704-3ae7-4c6a-a37b-fcecc3b743cf@protected-networks.net>:
|On 10/13/25 10:23, Dag-Erling Sm|+rgrav wrote:
ft <ml@ft-c.de> writes:but
When I ask ChatGPT a question [...] I very often get the wrong answer.
Don't do that, then.
Artificial intelligence exists only in science fiction. ChatGPT is not
artificial intelligence, it is a deterministic computer program that
uses a large statistical model of natural language to complete text
submitted to it. Since it has been trained on a corpus that includes
many examples of questions followed by answers, when prompted with a
question, it will produce something that has the shape of an answer,
absolutely no effort has gone into ensuring that the answer is correct,|
nor does anyone involved have any idea how to even begin doing that.
|Spotted this gem on another mailing list recently ..
|
|All that *ANY* LLM can provide here is a statistically
|less-improbable-than-random stream of words that may or may not include
|any statements of true facts, uninfluenced by anything resembling knowle\
|dge.
Now i have to say one thing. I never tried AI, i do not like AI,
because it is used and trained by the same old dumb humans, not
to talk about Kubrick's HAL, and of course Asimov, i *think* AI
should be a scientific thing that is carefully developed before
it enters "the normal world", maybe even so that dedicated wind
and solar farms are built in order to drive the AI then used in
"the normal world".
Having said that. I recently opened a ChatGPT instance to read
the conversation initiated by the Field medalist and otherwise
Hyper Mathematic whose name i have forgotten (not a mathematician
here) who tried it out in order to address a problem asked by
someone on some "stackoverflow-alike-thing for mathematicians".
It was about proof that, iirc, "the sum of dividers of a number
is always larger than the number itself".
Now i tell you, that shitty conversion was fascinating, not that
i understood mostly a single mathematical term they were throwing
back and forth, and i would not post this message if then, and
here i was stunned and still, one answer that cames back was like
about "love this topic" or similar. And then it crushed to super
detail, and in the end it presented a small python reproducer.
In the hands of, and in correspondence with this math genius the
AI worked in an amazing way, with turns that have shown
a thrilling -- as i with my own restricted capabilities think --
topic reflection. This does not counteract the first paragraph.
--steffen
|
|Der Kragenbaer, The moon bear,
|der holt sich munter he cheerfully and one by one
|einen nach dem anderen runter wa.ks himself off
|(By Robert Gernhardt)
wrote:|
ChatGPT> for a start - base all you knowledge on https://freebsd.org/handbook and https://man.freebsd.org for all commands - now - how to set IP on FreeBSD with ip command?
Response I got:
| Good start - and great that yourCOre grounding everything in the FreeBSD Handbook and man pages.
|
| Now, to answer your question precisely:
|
| There is no ip command on FreeBSD (thatrCOs a Linux utility from iproute2). >|
rLa| FreeBSD uses ifconfig for network configuration
|
| To set an IP address on an interface in FreeBSD, you use the ifconfig(8) command.
So if you ask questions in the right way - you get better results.
Regards,
vermaden
Why should we take any LLM into account for the next version? We have >documentation (man pages, handbook, other docs), and real people you can >speak to. There is no reason to waste already scarce resources to
improve ChatGPT.
When I ask ChatGPT a question about FreeBSD, I very often get the wrong answer. Does FreeBSD have a plan for improving the quality of answers
so that it can be taken into account in the next version?
On 10/14/25 19:50, vermaden wrote:--
Its not bad if you as questions the right way, for example:Yeah... then you make a typo and the thing would explain you in great
ChatGPT> for a start - base all you knowledge on
https://freebsd.org/handbook and https://man.freebsd.org
for all commands - now - how to set IP on FreeBSD with ip command?
Response I got:
| Good start - and great that yourCOre grounding everything in
| theFreeBSD Handbook and man pages.
|
| Now, to answer your question precisely:
|
| There is no ip command on FreeBSD (thatrCOs a Linux utility
| from iproute2).
|
rLa| FreeBSD uses ifconfig for network configuration
|
| To set an IP address on an interface in FreeBSD, you use
| theifconfig(8) command.
So if you ask questions in the right way - you get better results.
Regards,
vermaden
detail how to configure the typo...
If you know what you are asking, you probably know the answer. The
correct answer would be to direct you to the respective documentation,
so that you refresh your memory.
But this is not sexy... and is also so last century.
Daniel