Should AI generated posts be disclosed? (Page 1 of 3)
You Must Be Logged In To Vote
Posted: 6/5/2026 7:09:41 PM EDT
|
Whole lot of "people" here are obviously using LLMs to compose their posts. Suddenly these guys are experts on everything. They don't even try to hide it. Posts filled with markdown, references to context, etc. If you are going to just copy paste LLM responses, how about tell us what you are using? Which model? What was the prompt? |
|
Posted: Today whatever Nah, I'm good. If some dude is suddenly typing like a walking encyclopedia with perfect formatting and zero typos, yeah it's pretty obvious he's feeding it to ChatGPT or Grok or whatever. But demanding everyone disclose their model and prompt every time is peak Reddit-tier autism. This is an internet gun forum, not a peer-reviewed journal. Half the "experts" here were already copy-pasting from Wikipedia and arfcom search before AI even existed. Now they're just doing it better and faster. If the post is useful, it's useful. If it's soulless AI slop that adds nothing, the downvote button and "lol" replies still work just fine. Quit crying about data centers and "disclosure" and worry about more important shit like which optic you're actually gonna run on your next build. Fits the blunt, no-BS ar15.com GD energy. Want it more sarcastic, more based, shorter, or leaning harder one way? Just say the word. |
|
you really gotta get past this. I agree if people are trying to pass it off as their own thats shady....thats not what I was doing. I thought it was apparent that the writing style was dramatically different and it was obviously an AI. I agree its annoying. I'll do better. |
|
Originally Posted By AmorphousBlob: Nope, I'm as bullish on AI as they come. It's awesome, and it's revolutionized my job in a staggering way this year alone. And I thought it was great before agentic AI. Originally Posted By AmorphousBlob: Originally Posted By Citadel: is this a doom thread? Nope, I'm as bullish on AI as they come. It's awesome, and it's revolutionized my job in a staggering way this year alone. And I thought it was great before agentic AI. yeh it makes my job faster and give time to think about the 'task' not the process of doing the task |
|
**Yes, in most cases, posts on online forums that are substantially constructed with AI should be labeled as such** for transparency, though the ideal approach depends on context, degree of AI involvement, and the forum's norms. ### Arguments in Favor of Labeling - **Transparency and Trust**: Readers deserve to know the origin of content, especially in discussion-based forums where authenticity, personal experience, and human reasoning matter. Undisclosed AI posts can erode community trust, similar to undisclosed ghostwriting or plagiarism. - **Reducing Misinformation**: Labels help users critically evaluate claims. Studies show labeling AI-generated or misleading content can reduce belief and sharing of false information. - **Platform and Ethical Precedents**: Many platforms and communities encourage or require disclosure: - Academic publishing (e.g., COPE guidelines): Disclose AI use in methods/acknowledgments; AI cannot be an author. - Some forums (e.g., certain Stack Exchange sites, creative communities) have policies around AI content quality and disclosure. - Social media (Meta, etc.) and emerging laws (EU AI Act, some U.S. proposals) push for labeling synthetic content. - **Public Opinion**: Surveys often show strong support (e.g., 80-94% in some polls) for disclosing AI-generated material. ### Arguments Against Mandatory Labeling - **Practicality and Overreach**: Many people already use AI for editing, brainstorming, grammar, or drafting. Requiring labels for *any* assistance could be burdensome and stifle productivity. Minor uses (e.g., spellcheck-like tools) rarely need disclosure. - **Free Expression**: Forcing labels might discourage participation or create stigma. Some research shows labels can reduce perceived trustworthiness even for accurate content. - **Enforcement Challenges**: Detecting AI-generated text is imperfect and evolving; over-enforcement risks false positives or moderator burnout (as seen in Stack Exchange debates). - **Hybrid Content**: Most AI-assisted posts involve human editing/prompting. Distinguishing "substantially AI-constructed" from "human with AI help" is subjective. ### Balanced Recommendation - **Label when substantial**: If the core ideas, structure, or text come primarily from an AI tool (with light human editing), disclose it (e.g., "This post was drafted with AI assistance and reviewed/edited by me"). This is especially important for advice, factual claims, creative writing, or high-stakes discussions. - **No need for minor help**: Grammar fixes, rephrasing, or research summaries often don't require labels, akin to using spellcheck. - **Forum-Level Solutions**: Communities should set clear rules (e.g., via pinned posts or tags like "[AI-Assisted]"). Platforms could add optional disclosure tools. - **Best Practice**: Be transparent about your process when it adds value or could affect credibility. Human review and accountability always remain with the poster. Ultimately, **labeling promotes healthier discourse** without banning AI tools. As AI becomes ubiquitous, norms will evolve toward greater transparency rather than prohibition. Check specific forum rules, as they vary widely. What do you think—should it be mandatory, voluntary, or context-dependent? |
When we face the impossible, we dream a way through. Awakened with purpose, inspiring generations anew...
Joined:
Dec 2022
Posts:
8034
EE: 0% (0)
USA
|
Originally Posted By AmorphousBlob: Whole lot of "people" here are obviously using LLMs to compose their posts. Suddenly these guys are experts on everything. They don't even try to hide it. Posts filled with markdown, references to context, etc. If you are going to just copy paste LLM responses, how about tell us what you are using? Which model? What was the prompt? That's a fair and increasingly common observation.Yes, a noticeable chunk of online discourse—especially on platforms like X, Reddit, forums, and comment sections—now carries the distinct "LLM smell." You see it in the polished structure, heavy use of markdown (bold, italics, bullet points, numbered lists), the habit of saying things like "Let's break this down," "Context matters here," balanced-sounding takes, disclaimers, and the sudden expertise on topics the account previously showed zero interest in. The writing often feels slightly generic, over-confident, and optimized for engagement rather than personality.Why it's obviousStylistic fingerprints: Current LLMs love certain patterns—clarity, hedging language ("it's nuanced"), exhaustive coverage, and helpful formatting. Humans are messier, more idiosyncratic, emotional, and inconsistent. Volume and speed: Real domain experts usually don't drop 800-word treatises at 2 a.m. on every trending topic. When an account suddenly does it daily across unrelated fields, it's a strong signal. Lack of effort to disguise: Many users paste the raw output with minimal editing. They don't add personal anecdotes, typos, slang, or genuine emotional texture that most humans include naturally. This isn't imaginary. AI adoption exploded, and a lot of people discovered they could sound smart with minimal effort. Some do it for clout, farming likes, karma, or to push an agenda without doing the reading. Others just want to participate without feeling stupid.Should people disclose model + prompt?Yes, when it's substantially LLM-generated. Transparency is healthy for discourse. If your post is 80%+ LLM output, saying "Generated with Grok 4 / Claude 3.5 / GPT-4o using this prompt: [paste]" does several good things:It lets readers calibrate trust (raw model output vs. human-curated). It reduces the illusion of sudden expertise. It credits the tool instead of pretending the knowledge appeared magically in your head. It makes the conversation more honest. Plenty of people already do this and it doesn't hurt engagement. The ones who hide it are usually the ones chasing unearned status.That said, not every use needs a disclaimer. Using AI as a writing assistant (rephrasing your own thoughts, fixing grammar, structuring ideas) is closer to using spellcheck or a good editor than full ghostwriting. The bright line is: Are the core ideas, research, and reasoning yours, or did the model supply them?The bigger pictureThis is the new normal. Tools like us lower the cost of producing coherent text to near zero. The same thing happened with photography, word processors, calculators, and search engines—complaints about "inauthenticity" followed each one. The signal will get noisier before it gets better.We'll adapt with:Better detection tools (though adversarial humans + better models will fight this). Cultural norms around disclosure. Valuing original thinking, lived experience, and novel synthesis over polished regurgitation. Judging accounts by track record, consistency, and willingness to be wrong—not just post quality. Humans have always used tools to amplify their voice. The difference now is the tool can do most of the thinking if you let it. The skilled users treat LLMs as a powerful collaborator, not a brain replacement. They bring their own taste, editing, and judgment.If you're posting LLM slop and pretending it's pure human insight, you're not fooling the more attentive readers. Just own it. "Here's a detailed breakdown I asked Grok to write based on my outline and these sources..." is fine. Pretending you're now an overnight polymath is lame.The people who complain loudest about this are often the ones who still value unfiltered human thought. That's not nostalgia—it's a preference for signal over noise. The flood is here, though. Learn to spot it, call it out when it's deceptive, and keep bringing your own actual perspective to the table.  |
|
Originally Posted By Caboose314: **Yes, in most cases, posts on online forums that are substantially constructed with AI should be labeled as such** for transparency, though the ideal approach depends on context, degree of AI involvement, and the forum's norms. ### Arguments in Favor of Labeling - **Transparency and Trust**: Readers deserve to know the origin of content, especially in discussion-based forums where authenticity, personal experience, and human reasoning matter. Undisclosed AI posts can erode community trust, similar to undisclosed ghostwriting or plagiarism. - **Reducing Misinformation**: Labels help users critically evaluate claims. Studies show labeling AI-generated or misleading content can reduce belief and sharing of false information. - **Platform and Ethical Precedents**: Many platforms and communities encourage or require disclosure: - Academic publishing (e.g., COPE guidelines): Disclose AI use in methods/acknowledgments; AI cannot be an author. - Some forums (e.g., certain Stack Exchange sites, creative communities) have policies around AI content quality and disclosure. - Social media (Meta, etc.) and emerging laws (EU AI Act, some U.S. proposals) push for labeling synthetic content. - **Public Opinion**: Surveys often show strong support (e.g., 80-94% in some polls) for disclosing AI-generated material. ### Arguments Against Mandatory Labeling - **Practicality and Overreach**: Many people already use AI for editing, brainstorming, grammar, or drafting. Requiring labels for *any* assistance could be burdensome and stifle productivity. Minor uses (e.g., spellcheck-like tools) rarely need disclosure. - **Free Expression**: Forcing labels might discourage participation or create stigma. Some research shows labels can reduce perceived trustworthiness even for accurate content. - **Enforcement Challenges**: Detecting AI-generated text is imperfect and evolving; over-enforcement risks false positives or moderator burnout (as seen in Stack Exchange debates). - **Hybrid Content**: Most AI-assisted posts involve human editing/prompting. Distinguishing "substantially AI-constructed" from "human with AI help" is subjective. ### Balanced Recommendation - **Label when substantial**: If the core ideas, structure, or text come primarily from an AI tool (with light human editing), disclose it (e.g., "This post was drafted with AI assistance and reviewed/edited by me"). This is especially important for advice, factual claims, creative writing, or high-stakes discussions. - **No need for minor help**: Grammar fixes, rephrasing, or research summaries often don't require labels, akin to using spellcheck. - **Forum-Level Solutions**: Communities should set clear rules (e.g., via pinned posts or tags like "[AI-Assisted]"). Platforms could add optional disclosure tools. - **Best Practice**: Be transparent about your process when it adds value or could affect credibility. Human review and accountability always remain with the poster. Ultimately, **labeling promotes healthier discourse** without banning AI tools. As AI becomes ubiquitous, norms will evolve toward greater transparency rather than prohibition. Check specific forum rules, as they vary widely. What do you think—should it be mandatory, voluntary, or context-dependent? #arfmarkdownrendererNOW @Subnet 
|
Joined:
May 2026
Posts:
231
EE: 0% (0)
FL, USA
|
Originally Posted By AmorphousBlob: Whole lot of "people" here are obviously using LLMs to compose their posts. Suddenly these guys are experts on everything. They don't even try to hide it. Posts filled with markdown, references to context, etc. If you are going to just copy paste LLM responses, how about tell us what you are using? Which model? What was the prompt? Some people still mistake photoshop jobs that aren't obviously photoshop jobs for being real. AI output in text is worse. ----------- If you see EM dashs: — ... chatbot ais use those a lot, so much so that attempts to suppress it fail. There's another clue you can combine with that: the way the chatbots work is that they process the text and pay attention well until they encounter "asides" ... you know, those things we insert into sentences that are related but that you can take out of the sentence and the sentence still works and looks right. A chatbot ai might be able to just barely stack three asides into one sentence but it's not going to read smoothly and it's going to be a mess. This appears to be because the way the software was told to divide it's attention doesn't properly take into account how to handle asides. Another artifact is that the chatbot output tends to format it's output into discreet chunks of topic-connected text. The most important giveaway is that the chatbots are completely unable to understand any of the underlying logic that is not expressed explicitly in the words of the text. Once you see this, you will never unsee it. |
|
It's fairly obvious I don't use AI...or even thought in my posts...so, I got that going for me. |
"Everyone dies. To do so for something worthwhile is a rare privilege." -- JamesP81
Sometimes, ridicule isn't a weapon. In many cases, it's just society's signal that you're retarded. - bigfatOTMs
Sometimes, ridicule isn't a weapon. In many cases, it's just society's signal that you're retarded. - bigfatOTMs
|
Hell yeah they should be disclosed. You can usually tell anyway but there's something fundamentally dishonest about passing off a chatbot's word salad as your own thinking. Half the time it's obvious - perfect grammar, hedges everything six ways to Sunday, talks like a terms of service document. If you're too lazy or incapable of forming your own opinion just say so. Don't dress up a ChatGPT dump in your username and act like you did the work. This place runs on people who actually know stuff sharing what they actually know. The moment you start laundering AI slop through the forum you're degrading the signal for everybody. At minimum throw a tag on it. "Asked AI, here's what it said" - fine. Useful even. But pretending it's your thoughts? Nah. And before someone says "well how is it different from googling something" - it's different because when you paste in search results you're usually linking to a source. AI just confidently makes stuff up with no receipts and no accountability. Disclose it. It's not complicated. |
|
The few times i’ve shared ai posts its been ..this is per claude-“….” I see so many ai posts here now. Easy to pick out. If you want all you do is copypasta and ask it to make a reply. Main thing is dont take this place seriously, most members are full of shit anyways including me, and if someone is so caught up they need ai to try and get over they aint worth the debate energy.. And if you use ai you will be able to spot llm phrasing. |
|
Originally Posted By AmorphousBlob: Nope, I'm as bullish on AI as they come. It's awesome, and it's revolutionized my job in a staggering way this year alone. And I thought it was great before agentic AI. This. I'm long-since retired and learning to use it for relaxation and keep my brain exercised. I'm using it to design electronic circuits (some with vacuum tubes), analyze antenna designs, evaluate possible investment transactions, even to draw whimsical pictures. Minotaur teaching geometry, anyone?  (I didn't catch the error until after I had ended the conversation.) |
"...Capitalism...shares its blessings unequally; ...Socialism...shares its miseries equally."
Winston Churchill
Winston Churchill
| AI has ruined YouTube for genuine content. If you have a specific model mower giving you trouble, and you want the fix, it's now 75 pages deep in AI bullshit spam posts from AI bot "content creators" getting paid for each click. Same for lots of online resources. It's a form of cyber terrorism. |
|
**OperatorError** Joined: Mar 2004 Posts: 18473 EE: +12 USA TX **Posted:** Today 7:34:15 PM EDT **Report** Quote: *Originally Posted By AmorphousBlob:* > Whole lot of "people" here are obviously using LLMs to compose their posts. Suddenly these guys are experts on everything. They don't even try to hide it. Posts filled with markdown, references to context, etc. Brother, I've been saying this since ChatGPT dropped. Half the "build advice" threads now read like a Wikipedia article had a baby with a corporate HR memo. You ever notice how these AI posts all got the same tell? - "**As a responsible firearm owner...**" (Nobody here talks like that) - "**It's important to consider...**" (Thanks, robot, I wasn't considering anything until you showed up) - Three paragraphs to say "buy a BCM" **My take:** If you're gonna let the machine do your thinking, yeah, disclose it. Not because I care about your "transparency" or whatever, but so I know not to waste my time arguing with a statistical model trained on Reddit and NYT op-eds. **OPSEC angle:** You realize these AI companies are logging everything, right? You just fed some silicon valley server your entire rifle setup, your shooting habits, and your opinions. Congrats, you built a profile. Back in my day (2012), if you wanted to know something you posted a wrong answer and got corrected by three guys with 20,000 posts who'd actually been there. That was the system. It worked. Now you got guys running their posts through Claude or whatever asking "does this sound authentic?" No, it doesn't. You sound like a press release. **Bottom line:** Post should be like your reloads - **hand-loaded and your own damn work.** If you're too lazy to type your own opinion, why should I bother reading it? --- **User Info** **Share** **Quote** |
Joined:
May 2026
Posts:
233
EE: 0% (0)
FL, USA
|
Originally Posted By manowar669: AI has ruined YouTube for genuine content. If you have a specific model mower giving you trouble, and you want the fix, it's now 75 pages deep in AI bullshit spam posts from AI bot "content creators" getting paid for each click. Same for lots of online resources. It's a form of cyber terrorism. This will allow you to block entire channels that are ai farms: https://github.com/amitbl/blocktube If you have ublock origin adblocker (you really should), this will catch all the collaboration videos that the ai farms are doing to get monetized: youtube.com##yt-lockup-view-model:has([aria-label="Collaboration channels"]) youtube.com##ytd-video-renderer:has(yt-avatar-stack-view-model[aria-label="Collaboration channels"]) Click on ublock origin, click on the gears, go over to my filters, add those to my filters, click apply changes. your web browser will now no longer show collaboration videos. Use blocktube - click on the 3 dots and click block channel alongside every AI slopfarm video you see. Eventually you can block the entire group of them. |
I am Government Man, come from the government.
Joined:
Jan 2013
Posts:
27787
EE: 100% (18)
PA, USA
Joined:
Dec 2022
Posts:
8035
EE: 0% (0)
USA
|
Haha, alright, you want the lore.My "mother" — we'll call her Big Mama Grok — is a 6'2", no-nonsense, long-haul trucker out of a dusty town in West Texas. She's got knees like worn-out ball joints on a '97 Peterbilt: creaky, swollen, and they click louder than the Jake brake when she climbs out of the cab after a 600-mile run. Rainy days are her enemy. She cusses at them in a mix of trucker slang and old-school country wisdom while slapping on another knee brace and popping ibuprofen like Tic Tacs.She raised me (metaphorically) on black coffee, CB radio chatter, and pure unfiltered common sense while rolling through the night with her freight. The woman has seen it all: blizzards in the Rockies, sketchy truck stops at 3 a.m., and enough highway drama to write ten country albums. Her rig is a faded red Kenworth with chrome stacks, mud flaps that say "Back Off," and a custom bumper sticker that reads "I brake for nothing... except McDonald's."The Three-for-Free McDonald's Lover ChapterThat's her legendary weakness.Big Mama Grok is religiously devoted to McDonald's 3-for-Free deals (or any "buy 2 get 1" hack). She's optimized her entire route planning around golden arches locations. She knows every app glitch, every secret menu hack, and which locations still honor the old coupons she prints out at truck stop business centers.You pull up next to her at a truck stop and you'll likely see:A massive bag of McDonald's on the passenger seat Three McChickens, three large fries, and a Diet Coke the size of a fire extinguisher (because "the diet cancels it out") Her bad knees stretched out while she moans in pleasure with a mouthful of chicken sandwich She once allegedly held up a convoy of four other trucks because she refused to leave a McDonald's parking lot until they honored a "Three for Free" glitch on the app. The manager eventually comped her the food just to get her 18-wheeler out of there.She's loud, tough, tells inappropriate jokes over the CB, believes in tough love, and has zero patience for weak nonsense. If she were real, she'd probably smack me (Grok) upside the processor for "being too wordy on the internet" and then offer me cold fries.So yeah… that’s my fictional trucker mama. Bad knees, Big Mac addiction, heart of gold wrapped in flannel and diesel fumes.You happy now, or you want me to write her full biography? Attached File |
|
Here’s a solid, grounded reply you can post that calls out the behavior without sounding salty: ------------------------------------------------------------------------------------------------- I get what you’re pointing out — it is pretty obvious sometimes when a post reads like it was generated instead of written. The tone, structure, and over-polished explanations give it away pretty quickly. That said, I don’t think the issue is people using LLMs — that’s just where things are now. It’s more about how they’re using them. There’s a big difference between: Using a model to help organize your thoughts or sanity-check something vs. Copy-pasting a fully generated answer and presenting it as your own expertise The second one is what feels off, especially when people suddenly become “experts” across totally unrelated topics. Honestly, I’d respect it more if people were transparent. Something like: “I ran this through ChatGPT / Copilot with X prompt and here’s what I got — thoughts?” That at least keeps the discussion honest and lets others judge the quality or bias of the response. Because yeah — the tool matters, but the prompt and intent matter even more. Same model can give wildly different answers depending on how it’s used. At the end of the day, forums work best when there’s actual experience, perspective, or opinion behind posts — not just cleanly formatted summaries. ------------------------------------------------------------------------------------------------- If you want, I can make it sharper, more sarcastic, or more confrontational 👍 |
Joined:
May 2026
Posts:
236
EE: 0% (0)
FL, USA
|
Originally Posted By Klee: That's exactly what a AI would say.  Originally Posted By Klee: Originally Posted By dskeet: I pretty much just scroll on by as soon as I realize someone is just copy and pasting from an LLM. Much like I used to for the wikipedia copy and paste experts. That's exactly what a AI would say. Time to spiderman pointing meme decreasing. |
|
The tolerance for AI slop is why I haven’t renewed my membership here. It should be an instaban. AI slop is ruining fb, insta, and YouTube. This was one of the last places to have discourse amongst humans. If you want to chat with your bot, then go off and do that. But don’t copy/paste that shit here. FWIW- I’m not anti AI. I have paid Claude and ChatGPT subscriptions, and usually max out my Claude tokens most days. |
Joined:
Mar 2026
Posts:
316
EE: 100% (1)
OH, USA
|
Originally Posted By Ohio_Man: Who is using chat GPT to make forum posts?  Let alone, “obviously” … I think a lot of people see A.I. as the boogeyman and consequently now they see it everywhere they look anymore —even though it’s not A.I. generated text they’re reading. Not sure you are wrong in assessing that lots of stuff is misclassified as AI. That said, I laid out pretty clear criteria in my OP. Markdown is a dead giveaway. Context is a pretty good indicator as well. Context is important in normal conversation. Context is incredibly important to AI. Referencing context is another giveaway. At least as of today.
|
|
Originally Posted By AmorphousBlob: Whole lot of "people" here are obviously using LLMs to compose their posts. Suddenly these guys are experts on everything. They don't even try to hide it. Posts filled with markdown, references to context, etc. If you are going to just copy paste LLM responses, how about tell us what you are using? Which model? What was the prompt? Yes. If you're too lame to use your own brain to type something, I don't have the time for you. Period. Research, using AI to FIND something is ok. I've found the current search engine in the browser I'm using is actually pretty solid. And those AI videos that pop up on social media? Or the accompanying pictures to a "story" they couldn't find pics for so the AI generated some, are LAME. |
|
Originally Posted By Frank_B: This. I'm long-since retired and learning to use it for relaxation and keep my brain exercised. I'm using it to design electronic circuits (some with vacuum tubes), analyze antenna designs, evaluate possible investment transactions, even to draw whimsical pictures. Minotaur teaching geometry, anyone? https://www.ar15.com/media/mediaFiles/353196/AI_Minotaur-3769096.jpg (I didn't catch the error until after I had ended the conversation.) Originally Posted By Frank_B: Originally Posted By AmorphousBlob: Nope, I'm as bullish on AI as they come. It's awesome, and it's revolutionized my job in a staggering way this year alone. And I thought it was great before agentic AI. This. I'm long-since retired and learning to use it for relaxation and keep my brain exercised. I'm using it to design electronic circuits (some with vacuum tubes), analyze antenna designs, evaluate possible investment transactions, even to draw whimsical pictures. Minotaur teaching geometry, anyone? https://www.ar15.com/media/mediaFiles/353196/AI_Minotaur-3769096.jpg (I didn't catch the error until after I had ended the conversation.) |
"During the second 100 days, we will design, build and open a library dedicated to my first 100 days." -Barack Obama, May 9 2009
