Comment

Comment on It's rude to show AI output to people

brsrklf@jlai.lu ⁨4⁩ ⁨days⁩ ago

Every now and then I see a guy barging in a topic bringing nothing else than “I asked [some AI service] and here’s what it said”, followed by 3 paragraphs of AI-gened gibberish. And then when it’s not well received they just don’t seem to understand.

It’s baffling to me. Anyone can ask an AI. A lot of people specifically don’t, because they don’t want to battle with its output for an hour trying to sort out from where it got its information, whether it represented it well, or even whether it just hallucinated half of it.

And those guys come posting a wall of text they may or may not have read themselves, and then they have the gall to go “What’s the problem, is any of that wrong?”… Dude, the problem is you have no fucking idea if it’s wrong yourself, have nothing to back it up, and have only brought automated noise to the conversation.

source

Sort:hotnew top

expr@programming.dev ⁨3⁩ ⁨days⁩ ago
I was trying to help onboard a new lead engineer and I was working through debugging his caddy config on Slack. I’m clearly putting in effort to help him diagnose his issue and he posts “I asked chatgpt and it said these two lines need to be reversed”, which was completely false (caddy has a system for reordering directives) and honestly just straight up insulting. Fucking pissed me off. People need to stop brining AI slop into conversations. It isn’t welcome and can fuck right off.

The actual issue? He forgot to restart his development server. 😡

source
tias@discuss.tchncs.de ⁨4⁩ ⁨days⁩ ago

Dude, the problem is you have no fucking idea if it’s wrong yourself, have nothing to back it up

That’s not true. For starters you can evaluate it on its own merits to see if it makes logical sense - the AI can help solve a maths equation for you and you can see that it checks out without needing something else to back it up.

Second, agentic or multiple-step AI:s will dig out the sources for you so you can check them. It’s a smart search engine with no ads and better focus on the question asked.

source
- isVeryLoud@lemmy.ca ⁨3⁩ ⁨days⁩ ago
  Ok, I didn’t need you as a middle man to tell me what the LLM just hallucinated, I can do this myself.
  
  The point is that raw AI output provides absolutely no value to a conversation, and is thus noisy and rude.
  
  When we ask questions on a public forum, we’re looking to talk to people about their own experience and research through the lens of their own being and expertise. We’re all capable of prompting an AI agent. If we wanted AI answers, we’d prompt an AI agent.
  
  source
- Barrymore@sh.itjust.works ⁨4⁩ ⁨days⁩ ago
  And what happens when ~~mechahitler~~ the next version of Grok or whatever AI hosted by a large corporation that only has the interest of capital gains comes out with unannounced injected prompt poisoning that doesn’t produce quality output like you’ve been conditioned to expect?
  
  These AI are good if you have a general grasp of whatever you are trying to find, because you can easily pick out what you know to be true and what is obviously a ~~ridiculous mess of computer generated text that is no smarter than your phone keyboard word suggestions~~ AI hallucination.
  
  Trying to soak up all the information generated by AI in a topic without prior knowledge may easily end up with you not understanding anything more than you did before, and possibly give you unrealistic confidence that you know what is essentially misinformation. And just because an AI pulls up references, unless you do your due diligence to read those references for accuracy or authority on the subject, the AI may be hallucinating where it got the wrong information it’s giving you.
  
  source
  - Passerby6497@lemmy.world ⁨3⁩ ⁨days⁩ ago
    
    And just because an AI pulls up references, unless you do your due diligence to read those references for accuracy or authority on the subject, the AI may be hallucinating where it got the wrong information it’s giving you.
    
    This. I’ve had the AI provide me vendor documentation that said the opposite of what it says the doc says.
    
    source
- Mirodir@discuss.tchncs.de ⁨4⁩ ⁨days⁩ ago
  On the second part. That is only half true. Yes, there are LLMs out there that search the internet and summarize and reference some websites they find.
  
  However, it is not rare that they add their own “info” to it, even though it’s not in the given source at all. If you use it to get sources and then read those instead, sure. But the output of the LLM itself should still be taken with a HUGE grain of salt and not be relied on at all if it’s critical, even if it puts a nice citation.
  
  source
- tomalley8342@lemmy.world ⁨3⁩ ⁨days⁩ ago
  If you have evaluated the statement for its correctness and relevance, then you can just own up to the statement yourself. There is no need to defer responsibility by prefacing it with “I asked [some AI service] and here’s what it said”. That is the point of the article that is being discussed, if you’d like to give it a read sometime.
  
  source
- brsrklf@jlai.lu ⁨3⁩ ⁨days⁩ ago
  I am speaking from experience.
  
  The latest example of that I encountered had a blatant logical inconsistency in its summary, a CVE that wasn’t relevant to what was discussed, because it was corrected years before the technology existed. Someone pointed to it.
  
  The poster hadn’t done the slightest to check what they posted, they just regurgitated it. It’s not the reader’s job to check the crap you’ve posted without the slightest effort.
  
  source
- bdonvr@thelemmy.club ⁨3⁩ ⁨days⁩ ago
  It gives you some links but in my experience what it says in the summary isn’t always the same as what’s in the link…
  
  source
- SparroHawc@lemmy.zip ⁨3⁩ ⁨days⁩ ago
  
  with no ads
  
  For now.
  
  Eventually it becomes a search engine that replaces the ads on the source material with its own ads, thus choking out the source’s funding and taking it for itself.
  
  source
- MousePotatoDoesStuff@lemmy.world ⁨2⁩ ⁨days⁩ ago
  “With no ads” Google used to have no ads. And especially with how much it cost to run even today’s LLMs, let alone tomorrow’s ones… enshittification is only a matter of time.
  
  source
- michaelmrose@lemmy.world ⁨3⁩ ⁨days⁩ ago
  No. They don’t actually use sources and can’t tell you why they said what they said. This actually reverses cause and effect the source is part of the inference and since it didn’t and in fact can’t go and read them frequently contains both imaginary sources and sources which don’t actually support the assertion.
  
  The ability to evaluate the source requires the same skills and information as the production of the original text would have required which means perforce if you use chatGPT to produce text which you lack the skills to produce you also lack the skills to evaluate it.
  
  You COULD go and read primary sources and in turn evaluate it but at that point its not that quick.
  
  source