Re: [GLLUG] Using LLM for support answers - please don't (Was Re: British Gas DKIM failure?)
On 2024-01-28 14:06, Carles Pina i Estany via GLLUG wrote: > I am normally not active on this forum, but read most of the messages > and do have some knowledge in linux. For me to state that a policy from > another site, just because it makes sense, should also retrospectively > appy here just because you think it makes sense, is a wrong assumption. > If you think that policy should apply here, lets discuss and agree to that. > Not just retrospectively apply it to a person, that for all intents and > purposes is trying to help Hendrik is not conductive to free and open > discussions. > For me the usage of LLMs, if done correctly, is a great tool. As any tool it > has its limitation, the famous hallucinations to name but one. I managed development of an AI project between 1995 and 2000. In my role as "speaker to suits" I had to explain it to people` with zero understanding of IT. The more I came to understand it the less I liked it. For our project it was fine because our requirement was for credibility and not accuracy. We are currently at the "unrealistic expectations" phase of the hype-cycle and blind acceptance is rife.. The problem is that it generates plausible answers and not necessarily correct ones. In situations where correctness is important the error-rate needs to be watched. If users are accustomed to blindly accepting whatever the computer says 99.99% right might be worse than 90%. My advice is that it's often a mistake to use LLM for any job that you can't do better yourself. > Let's try to be nice to eachother, especially when somebody is doing > his/her/its best to help Jan Indeed. Although we don't necessarily have to agree. I would prefer that LLM generated material should always be flagged as such. I'm unsure whether I should go further and consider discussions in which LLM-generated arguments have been presented as 'tainted.' -- Bernard Peek b...@shrdlu.com -- GLLUG mailing list GLLUG@mailman.lug.org.uk https://mailman.lug.org.uk/mailman/listinfo/gllug
Re: [GLLUG] Using LLM for support answers - please don't (Was Re: British Gas DKIM failure?)
Hello, On Sun, Jan 28, 2024 at 02:37:45PM +, Jan van Bergen via GLLUG wrote: > How Carles used it, with giving a reference stating that the lines in > question came from a LLM, while still making sure that the info is correct > to me is very much how you should use tools like this, and I had absolutely > no issue with it. My issues with it are pretty much the same as StackOverflow's issues with it. Any of us could have done the same, including Henrik themselves. Do we want support venues that are just people pasting ChatGPT to each other, web searches pulling back hits that are just more of that? We have to spot that it's from an LLM and check the reference ourselves. We don't know whether Carles did that for us. We can't generally trust the LLM user to do that. Carles could have asked the LLM the question, done the research themselves to check that what the LLM came back with is correct, and then written a response that they believe to be true and factual, in which case that's fine. But we don't know that happened because it's just a paste from ChatGPT. > Maybe you're a language virtuoso and don't need tools to write, > not everybody is like that. Nice personal attack noted, but we aren't talking about writing prose. > Let's try to be nice to eachother, especially when somebody is doing > his/her/its best to help I think my request was politely phrased and backed up with good reasoning, whether you agree with the reasoning or not. I don't think that pasting ChatGPT responses is someone doing their best to help people. Thanks, Andy -- https://bitfolk.com/ -- No-nonsense VPS hosting -- GLLUG mailing list GLLUG@mailman.lug.org.uk https://mailman.lug.org.uk/mailman/listinfo/gllug
Re: [GLLUG] Using LLM for support answers - please don't (Was Re: British Gas DKIM failure?)
On 28/01/2024 14:37, Jan van Bergen via GLLUG wrote: Let's try to be nice to each other, especially when somebody is doing his/her/its best to help +1 -- GLLUG mailing list GLLUG@mailman.lug.org.uk https://mailman.lug.org.uk/mailman/listinfo/gllug
Re: [GLLUG] Using LLM for support answers - please don't (Was Re: British Gas DKIM failure?)
On 2024-01-28 14:06, Carles Pina i Estany via GLLUG wrote: Hi, On 28 Jan 2024 at 13:23:26, Andy Smith via GLLUG wrote: Hello, On Sun, Jan 28, 2024 at 12:42:20AM +, Carles Pina i Estany via GLLUG wrote: > (this is a copy-paste from a... ChatGPT conversation): Please don't. To clarify, that was only the list of things that could have been wrong on why opendkim reported "bad signature". To my knowledge, the list seems correct and can be usefl. I am not a professional mail sysadmin (even though I set up email servers, during years, in different environments). The rest of the email is hand typed and brain thought! Anyway, I'll not do it again. If this was a StackOverflow site, your response would not be permitted because you used an LLM (ChatGPT). yes, but this is not StackOverflow (so didn't think that adding 4 lines that I thought that were well explained) was a problem. Stating the source. I thought that it was a good description to help Henrik that could have happened and fix the issue. I think that StackOverflow's reasoning for their policy is sound and would apply here also: https://meta.stackoverflow.com/questions/421831/temporary-policy-generative-ai-e-g-chatgpt-is-banned In a nutshell, any of us, including Henrik, can easily use an LLM yet what we can't easily do without domain knowledge is tell when an LLM is *incorrect*. I do have some (limited) domain knowledge and I thought that the 4 lines were quite correct and a good summary (else, I wouldn't go answering things that I have no idea). I might still be wrong, in this case I apologise and hope to learn. When someone asks a question on a mailing list like this, I'd like to think their question would be given as much respect as if it were asked on a Stack site. This is my third email trying to help Henrik, including sharing some scripts that I use for a similar case. I really only want to help Henrik, and I used tools that I had in hand to try to explain one of the errors. I will not do it another time. Sorry for the confusion here! I am normally not active on this forum, but read most of the messages and do have some knowledge in linux. For me to state that a policy from another site, just because it makes sense, should also retrospectively appy here just because you think it makes sense, is a wrong assumption. If you think that policy should apply here, lets discuss and agree to that. Not just retrospectively apply it to a person, that for all intents and purposes is trying to help Hendrik is not conductive to free and open discussions. For me the usage of LLMs, if done correctly, is a great tool. As any tool it has its limitation, the famous hallucinations to name but one. However many times already I have used it to create an outline of an article or presentation. Even if you're an expert in the area, just getting past a blank page can be hard. As long as you check the outcome and make sure that any mistakes are corrected and missing info is added it can be an extremely valuable tool in improving productivity. I am actively encouraging all the developers in my department to use it. How Carles used it, with giving a reference stating that the lines in question came from a LLM, while still making sure that the info is correct to me is very much how you should use tools like this, and I had absolutely no issue with it. I often ask LLMs to rewrite paragraphs I have written to make them easier to read, as English is not my first language. Obviously I would make sure I still agree with what is said and often it is a dialogue with the LLM till I am happy with the proposed text. Maybe you're a language virtuoso and don't need tools to write, not everybody is like that. Let's try to be nice to eachother, especially when somebody is doing his/her/its best to help Jan -- GLLUG mailing list GLLUG@mailman.lug.org.uk https://mailman.lug.org.uk/mailman/listinfo/gllug
Re: [GLLUG] Using LLM for support answers - please don't (Was Re: British Gas DKIM failure?)
Hi, On 28 Jan 2024 at 13:23:26, Andy Smith via GLLUG wrote: > Hello, > > On Sun, Jan 28, 2024 at 12:42:20AM +, Carles Pina i Estany via GLLUG > wrote: > > (this is a copy-paste from a... ChatGPT conversation): > > Please don't. To clarify, that was only the list of things that could have been wrong on why opendkim reported "bad signature". To my knowledge, the list seems correct and can be usefl. I am not a professional mail sysadmin (even though I set up email servers, during years, in different environments). The rest of the email is hand typed and brain thought! Anyway, I'll not do it again. > If this was a StackOverflow site, your response would not be > permitted because you used an LLM (ChatGPT). yes, but this is not StackOverflow (so didn't think that adding 4 lines that I thought that were well explained) was a problem. Stating the source. I thought that it was a good description to help Henrik that could have happened and fix the issue. > I think that StackOverflow's reasoning for their policy is sound and > would apply here also: > > > https://meta.stackoverflow.com/questions/421831/temporary-policy-generative-ai-e-g-chatgpt-is-banned > > In a nutshell, any of us, including Henrik, can easily use an LLM > yet what we can't easily do without domain knowledge is tell when an > LLM is *incorrect*. I do have some (limited) domain knowledge and I thought that the 4 lines were quite correct and a good summary (else, I wouldn't go answering things that I have no idea). I might still be wrong, in this case I apologise and hope to learn. > When someone asks a question on a mailing list like this, I'd like > to think their question would be given as much respect as if it were > asked on a Stack site. This is my third email trying to help Henrik, including sharing some scripts that I use for a similar case. I really only want to help Henrik, and I used tools that I had in hand to try to explain one of the errors. I will not do it another time. Sorry for the confusion here! -- Carles Pina i Estany https://carles.pina.cat signature.asc Description: PGP signature -- GLLUG mailing list GLLUG@mailman.lug.org.uk https://mailman.lug.org.uk/mailman/listinfo/gllug
[GLLUG] Using LLM for support answers - please don't (Was Re: British Gas DKIM failure?)
Hello, On Sun, Jan 28, 2024 at 12:42:20AM +, Carles Pina i Estany via GLLUG wrote: > (this is a copy-paste from a... ChatGPT conversation): Please don't. If this was a StackOverflow site, your response would not be permitted because you used an LLM (ChatGPT). I think that StackOverflow's reasoning for their policy is sound and would apply here also: https://meta.stackoverflow.com/questions/421831/temporary-policy-generative-ai-e-g-chatgpt-is-banned In a nutshell, any of us, including Henrik, can easily use an LLM yet what we can't easily do without domain knowledge is tell when an LLM is *incorrect*. When someone asks a question on a mailing list like this, I'd like to think their question would be given as much respect as if it were asked on a Stack site. Thanks, Andy -- https://bitfolk.com/ -- No-nonsense VPS hosting -- GLLUG mailing list GLLUG@mailman.lug.org.uk https://mailman.lug.org.uk/mailman/listinfo/gllug