Oct 29, 2026, 1:01 PMit is never appropriate to write "i asked an llm and this is what it said:"
either the listener uses llms, in which case they can use an llm on their side, or they don't want to use llms, in which case inflicting llm text on them is rude and inconsiderate
:neobot_drake_dislike: I asked an LLM
:neobot_drake_like: I asked my cat
@wingo llm said that you are right
@wingo “was my original question ‘hey, can somebody ask an LLM about this and get back to me?’ No? Then sit down. Take a seat. Take several seats.”
@wingo folks want to post it so they get clout if it's right, but to mention that it's ai so they can be removed from the consequences of it being wrong. which is a funny exemplar of the entire scam
Oct 29, 2026, 1:06 PM@wingo@mastodon.social i demand the prompt, I want their input. If their not willing to tell me exactly what they told 5 corporations in a trenchcoat, then fuck off with their 'help'...
@wingo Don't be a meat proxy.
https://meatproxy.me/
Oct 29, 2026, 4:38 PM@billgateseatsbrains re: permissions, of course (edit: with regards to original toot, not pablo’s reply ofc) @p4bl0
@billgateseatsbrains @wingo It's not mine, but I believe it exists precisely to be shared.
Oct 29, 2026, 1:19 PM@wingo the meat proxy version of Let Me Google It For You
@ebassi @wingo : The Meat Proxy "I asked an LLM your question and it said 'X'." is much worse than the "Let me Google that for you." thing. I've done "LMGTFY" for many years, referring people to the site https://lmgtfy.com/ . And the difference is that LMGTFY is a way of *showing them how to Google it themselves*, so that *next time they can Google it themselves.*
By contrast, the Meat Proxy "I asked an LLM" thing is not even showing you the prompt or identifying the LLM.
@wingo ah but the guy wanted to regale you with the answer because he's a prompt wizard so managed to get a much better answer from the llm than "regular" folks
@wingo I love getting not working code in my PRs from my teamlead, that actually is worse than my solution.
"I just asked ChatGPT how to fo this better and ..." not even took a minute to paste it in and see if the tests even passed (they never do)
I love replying with a screenshot of a LLM stating the exact opposite,
@wingo actually I think it's much more considerate (and honest) that they let me know it's from an llm vs just passing it off as their own work
Much of the time, the original source is obvious. Common traits of AI answers:
* Everything is spelled and punctuated immaculately.
* Answers are often rather verbose and long.
* Answer includes a bullet list or other kind of list, usually three to five points long. (Yes, I do see the irony of using a short bullet list in my own comment. LOL)
* The wording sounds far too mature and/or educated for the usual writing style and level of the person posting the information.
* The answer, when checked, turns out to be wrong.
@wingo A friend twice texted me slop info from google's gobbledygook machine. I deleted his msgs unread, and with no response. I was doubly pissed, because he knows how much I detest google, as well as so-called "AI". He seems to have gotten the idea, because there hasn't been a third time.
Oct 29, 2026, 2:01 PM@wingo oh wow, a hit tweet
Oct 29, 2026, 2:02 PM@cfbolz seems to have hit a nerve, i don't even hate llms this much
Oct 29, 2026, 2:16 PM@wingo it's not even about whether you like LLMs or not. it's still a very valid rule simply for communication reasons
@wingo it makes you sound like a moron is the main issue. ive hit countless people with "bro asked ai"
@wingo I already blocked few people here because of that...
Oct 29, 2026, 2:36 PMsomebody did this right here on mastodon earlier this year in a reply to me
oh how i roasted them
@wingo if someone is stupid enough to rely on a LLM for their reply it's always useful to know this upfront, so you can stop at that point.
Of course it would still be better for them to wise up and not to do it in the first place.
@wingo
Indeed , some people swing the hammer , other people let the hammer swing them . . .
@wingo
Those late night psychic ripoff ads keep coming to mind. 'I asked chatgpt' automatically translates to 'I asked Miss Cleo' and I stop listening.
I have experienced worse: I asked in a local group where to buy something … and the reply was that I can ask ChatGPT, it would know … and they did not even copy and paste the reply.
@wingo Plausible-sounding but wrong answers are actually harmful in this situation, because I have to waste time debunking them. If you can't answer from your own knowledge, please do not respond.
@wingo and it should always end by saying “it’s the least I could do” because it sure the hell is.
@wingo I asked LLM and it said you are absolutely right.
@wingo Telling other people what an LLM said is like forwarding spam.
Oct 29, 2026, 7:37 PM@wingo
I’m totally fine when someone does that because it’s my clue to stop reading and move on to the next thing 🙃
It is appropriate, if the LLM's text is followed by a thorough analysis and, if necessary, debunking. It's fun to do that to boosters. Even read about a professor doing that: demanding their students generate an essay and then disprove it.
@wingo If you are speaking to a subject matter expert about their field of expertise it's downright disrespectful.
@wingo At least they acknowledge using an LLM, rather than plagiarizing the output without attribution. 😀
@wingo at the very least, in polite society, this should be flanked by "I checked the answer and it holds up".