Can I get permission to create an AI user? (Page 1 of 2)
Posted: 8/25/2025 10:25:25 PM EDT
|
Simple idea: I'd like to create an AI agent that has its own Arfcom account, and responds only when tagged in a thread. I'm definitely not wanting to break the rules :). My thought is that it would end up being used in much the same way Grok is used on X (i.e., "@grok, is this real?") I should be able to build it entirely on my end, without any changes or feedback from staff. I'll incur any associated costs. If it turns out to be something you'd like to make a real part of the site, I'll be happy to provide both the code and assistance setting it up. If it turns out to be something you hate, I'll turn it off - no harm done. What sayeth site staff? |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
A coworker did that. Set the personality as an urban youth. turned it loose in a chat during a live webcast of his kids homeschooling program. it asked uncomfortable personal questions of the teacher lady totally unrelated to the topic at hand. it really got fun when he popped in another of similar background but of the female of the species. AI "M" started courting AI "F" using the proper terminology and language skills appropriate for its personality |
Will not shelter in place
|
Originally Posted By sauerkraut: Permission granted |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
Originally Posted By backbencher: The amount of dick pics it will receive will be staggering. |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
Originally Posted By right_rudder: A coworker did that. Set the personality as an urban youth. turned it loose in a chat during a live webcast of his kids homeschooling program. it asked uncomfortable personal questions of the teacher lady totally unrelated to the topic at hand. it really got fun when he popped in another of similar background but of the female of the species. AI "M" started courting AI "F" using the proper terminology and language skills appropriate for its personality That's interesting, kind of neat, weird and definitely scary. "This business will get out of hand..." |
I love a good double entendre.
|
Please, don't Look at how enshittified other social media has become with both AI generated content submissions and fake comments to drive engagement (eg Reddit, Twitter) There's already people copy-pasting AI garbage on here in attempts to seem smart Also, your bot will face the same fate as every other 
|
"Rattlecan that fu**er and take her to McDonalds for dinner" -veritas8985
|
Originally Posted By giantpune: All the AI crap needs to go in its own subforum. Along with all the users who read a question in a thread, then feed it to chatGPT, and come back here to copy/paste a 87,000 word essay into a thread response. Nobody is reading that crap. This. Anytime a post starts with "Grok says:" or "ChatGPT says:" I immediately close the thread. I don't care what AI has to say about ARFCOM topics, I want to know what real people have to say. |
|
Originally Posted By m200maker: It could prove to be helpful in the tech forums. But it would likely go south in GD the first day. Totally different use case, and would ideally require a fair amount of historical post data to do well - but if given access to the post archive (by scraping, or preferably, by API or DB dump) I could create a RAG with the combined knowledge of an entire tech forum at its fingertips. Actually... if I could get someone from staff to DM me to discuss details, I bet I could put together a much superior search function. We have a lot of data, but not that much. If our current site search is Google, I could deliver Perplexity. |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
Originally Posted By Striker: @Stylz I'm going to go out on a limb and say no but that is outside my wheelhouse. I'm intentionally not looking for a lot of people to know about it, especially at this point. Note also that I'm happy to comply with whatever rules staff feel are appropriate. I'm absolutely not proposing a bot that jumps into threads without being called or that posts things on its own - especially as an experiment, it should be as small and as contained as possible. My first pass at a ruleset: * the CoC is included in its entirety in the bot's prompt * have a dedicated user account, so it's 100% clear that it's not a human - I'd be willing to pay for it to have a Bronze account at the very least * only responds when directly mentioned using the "@" syntax - I'd have it triggered by the notification sent by email from the account * limited to one or more specific subforums, especially at first. It can be broadened later, if it's actually useful I would not put it in GD, at least not until I was confident that it worked well, followed the rules, and wasn't overly disruptive. I do think it would be a very interesting addition for GD in particular... but that's not where it should start. ETA: I can also add a blacklist of situations where it won't respond. For example, if "@grok, summarize this thread" becomes an issue, I could include a rule that prevents it from doing that. |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
Originally Posted By SimonPhoto: I get it, that's why I wanted to ask first. I've asked similar things in DMs before and not gotten a response, so this seemed like the next best thing. I'm intentionally not looking for a lot of people to know about it, especially at this point. Note also that I'm happy to comply with whatever rules staff feel are appropriate. I'm absolutely not proposing a bot that jumps into threads without being called or that posts things on its own - especially as an experiment, it should be as small and as contained as possible. My first pass at a ruleset: * the CoC is included in its entirety in the bot's prompt * have a dedicated user account, so it's 100% clear that it's not a human - I'd be willing to pay for it to have a Bronze account at the very least * only responds when directly mentioned using the "@" syntax - I'd have it triggered by the notification sent by email from the account * limited to one or more specific subforums, especially at first. It can be broadened later, if it's actually useful I would not put it in GD, at least not until I was confident that it worked well, followed the rules, and wasn't overly disruptive. I do think it would be a very interesting addition for GD in particular... but that's not where it should start. ETA: I can also add a blacklist of situations where it won't respond. For example, if "@grok, summarize this thread" becomes an issue, I could include a rule that prevents it from doing that. Coming soon, arfcom gets its own shitty "clippy" AI chatbot floating box down in the corner of the page to annoy everybody. 
|
|
So now you want to toss some more AI bots into the mix? What makes you think they aren't here already? What would the point in all of that be? Just to stimulate site traffic? To track replies and data from different users to develop it's own database on everyone? Seems like a weird request...... 
|
| How 'bout we not do that. I have to sift through enough ai-generated crap at work. Coming here is a leisure activity. If I start having to deal w/ ai-generated content here, then my best option becomes not coming here. |
Nowhere in the Constitution will you find the phrase, "unless it's inconvenient to Government".
|
Originally Posted By jcatcg: How 'bout we not do that. I have to sift through enough ai-generated crap at work. Coming here is a leisure activity. If I start having to deal w/ ai-generated content here, then my best option becomes not coming here. This guy talks the truth...... AI members have already infected FB posts, some web forums, and everything else online. If I come here and have to start figuring out if members are real people, or just AI bots.....that's just one more reason to stop surfing ARFCOM. And as far as using it to "help out" in the tech forums.....then why even have a "forum"? It would basically just turn into a google search for gun tech. |
|
Originally Posted By imq707s: So now you want to toss some more AI bots into the mix? What makes you think they aren't here already? What would the point in all of that be? Just to stimulate site traffic? To track replies and data from different users to develop it's own database on everyone? Seems like a weird request...... The point? I'm primarily motivated by curiosity at this point. I think it would be useful for technical topics, and that it would be interesting from a social perspective to see how it's used in this particular community - but the more I think about it, the more actually usefulness I think it could be. It would absolutely not be to simulate traffic. I want to explicitly label it as what it is. I don't want it to proactively post anything; it should only make an appearance when specifically requested. I want people to be able to easily ignore it if they don't want to see it. A dedicated account with an obvious robot avatar and an appropriate account name would go a long way toward all of that. If you, as a user, don't want to see its posts - blacklist its account. If it violates the CoC or is otherwise harmful to the site - lock the account. To be clear here - I'm not asking for special treatment, I see nothing in the CoC that would prohibit this. The CoC explicitly says that multiple accounts are allowed by permission. I'm just following the process the best I can :) |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
Originally Posted By AR_Dale: The AI bot on Transit forum is so hated there was a several page thread of nothing but users bitching at Admin to get rid of it. If it causes an issue, staff can lock the account and effectively turn it off immediately. If it's successful and we decide to expand it to all forums at some point, any subforum mod can ban it from their subforum. I can't think of a better way to test the water for something like this - and we should be testing the waters. You can't leave the future to everyone else and expect to stay relevant. |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
Originally Posted By imq707s: This guy talks the truth...... AI members have already infected FB posts, some web forums, and everything else online. If I come here and have to start figuring out if members are real people, or just AI bots.....that's just one more reason to stop surfing ARFCOM. And as far as using it to "help out" in the tech forums.....then why even have a "forum"? It would basically just turn into a google search for gun tech. |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
Originally Posted By imq707s: To track replies and data from different users to develop it's own database on everyone? Absolutely not. While it would be easiest to implement via something like an OpenAI Pro account (which I have), I could also implement it using my own, local inference. As a simple chatbot, I was planning on explicitly not giving it any memory beyond the thread it's mentioned in. Now that we're talking about it possibly being a RAG with access to historical data in tech forums, that would be something I'd want to work closely with staff on to make sure I'm not associating content with posters at all. I'd work in the open, too. You could easily see the code I'm using, fork it for your own purposes, etc. I'm not wanting to do anything nefarious or shady whatsoever - that's why I'm asking for explicit permission. |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
Originally Posted By SimonPhoto: Like I said, that's why I'm asking to create an account for it. If you don't like it, graybar it. If it causes an issue, staff can lock the account and effectively turn it off immediately. If it's successful and we decide to expand it to all forums at some point, any subforum mod can ban it from their subforum. I can't think of a better way to test the water for something like this - and we should be testing the waters. You can't leave the future to everyone else and expect to stay relevant. So...how much data would your AI Bot collect on users? People post up a lot of information here....where they live, political views, what they do with their kids on the weekends, how many kids they have, where they work, what guns they own, what cars they drive....tons and tons of info, that I'm sure would be easily filed away in a database for later use. Not saying your bot would be developed for nefarious purposes......but I would think that if wanted, it could be used to collect and record all kinds of personal information on users......and I would think that in itself would scare a lot of people away from ARFCOM. Or am I just overthinking things? EDIT: I see that you mentioned this in the above post..... |
|
Originally Posted By DADGAD: If everyone knew it was a bot, it would be one of the more entertaining shit shows. Maybe even if we didn't. I might do it on another forum, though, because I am very curious how long it would take people to realize it wasn't human - or if they ever did. I'm about 90% certain I could get it to fly under the radar entirely if that were the intent. |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
Originally Posted By imq707s: So...how much data would your AI Bot collect on users? People post up a lot of information here....where they live, political views, what they do with their kids on the weekends, how many kids they have, where they work, what guns they own, what cars they drive....tons and tons of info, that I'm sure would be easily filed away in a database for later use. Not saying your bot would be developed for nefarious purposes......but I would think that if wanted, it could be used to collect and record all kinds of personal information on users......and I would think that in itself would scare a lot of people away from ARFCOM. Or am I just overthinking things? EDIT: I see that you mentioned this in the above post..... I work in healthcare, and am extremely sensitive to data privacy. I wouldn't architect it in a way that it would be aware of anything that a normal user would be aware of - and would specifically never retain any information that could be useful to an adversary trying to harm our community. That means it would be at least as restrictive as our archives policies, and likely much more so. I see no reason to keep any poster-specific information longer than is necessary to respond to a single thread. |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
Originally Posted By Advance: As mentioned above... Looking at this through a different lens, I could see a lot of harvested data ending up in the hands of those who vehemently oppose 2A. How many big data orgs are on our side? But I'll re-iterate again: I'm not currently thinking about retaining any data whatsoever. If it ends up going in a direction that required it, I would work very closely with site staff and not create anything that would be a viable target for data exfiltration - be it via hack or via subpeona. I'm an anarchist for God's sake. I'm not going to make it easier to further curtain our rights. |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
If you run it on the tech forums, its just gonna hallucinate. Spit out incorrect historical facts. Or tell noobs the incorrect way to assemble their AR. AI fanbois are about like people when they get a 3D printer. They try to shoehorn AI/3D printing into every aspect of everything. Just stop it. Is nothing sacred? We don't need it. Most people don't want it. Its almost like a new skin on a website I saw recently.
|
|
Originally Posted By giantpune: If you run it on the tech forums, its just gonna hallucinate. Spit out incorrect historical facts. Or tell noobs the incorrect way to assemble their AR. AI fanbois are about like people when they get a 3D printer. They try to shoehorn AI/3D printing into every aspect of everything. Just stop it. Is nothing sacred? We don't need it. Most people don't want it. Its almost like a new skin on a website I saw recently. If there are specific concerns like "tell noobs the incorrect way to assemble their AR", I can and will add specific rules to prevent it from responding in those contexts. As for being a "fanboi"... well, that's a great example. I have a Bambu X1C sitting next to me. I 3D print a ton of things, but it's all things that are actually useful and that solve real problems. For example, a while back I designed and printed an adapter to let my physically mount my PVS-14 to a professional camera lens so I can take high-quality stills and video with it. I'm currently working on a support bracket to reinforce the headlight switch assembly for 88-93 GMT400 pickups that will prevent it from pushing back into the dash when the four tiny screws break the plastic they're attached to. There's a big difference between "chasing the shiny things" and "deeply learning and using new tools". I'm definitely in the latter camp. |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
Originally Posted By SimonPhoto: If there are specific concerns like "tell noobs the incorrect way to assemble their AR", I can and will add specific rules to prevent it from responding in those contexts. The specific concern is that people are tired of AI getting shoehorned into everything. This isn't for the benefit of the users. It is for the amusement of one guy who writes AI bots for a living. Figure out a way to do it that doesn't obtrude into or alter the browsing experience of everybody else who is tired of half baked AI every time we turn around. |
|
Originally Posted By giantpune: The specific concern is that people are tired of AI getting shoehorned into everything. This isn't for the benefit of the users. It is for the amusement of one guy who writes AI bots for a living. Figure out a way to do it that doesn't obtrude into or alter the browsing experience of everybody else who is tired of half baked AI every time we turn around. Does "only speaks when spoken to" not satisfy this? |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
Originally Posted By SimonPhoto: I understand that criticism, and did my best to address it above. Does "only speaks when spoken to" not satisfy this? What are YOU hoping to gain from this? "I'm primarily motivated by curiosity at this point"......so what are your secondary motivations? Not trying to be a jerk...just truly wondering why someone would put in the time and effort to do this.......for no reason other than "I want to see what happens". Seems like a lot of investment....for not much return.... |
|
Originally Posted By SimonPhoto: Does "only speaks when spoken to" not satisfy this? No. In your vision, one person speaks to the thing, and it replies back for the other 4500 users to have to see. Your idea for "oh, you can just ignore it" defaults to opting everybody in, which is backwards. And even the people who do ignore it will still see gray bars and quoted posts as other users talk to it. It is still intrusive to everybody who doesn't want anything to do with it. It would be better if it only existed via the PM/email system. The only users who ever saw anything it had to say were the ones who PM'd it first, and it responded back in a way that only they could see. If it posts anything anywhere that a user who hasn't asked for its input can see it, it is too intrusive. |
ARFCOM’s CHO
Joined:
Aug 2022
Posts:
3482
EE: 100% (1)
FL, USA
|
Originally Posted By imq707s: What are YOU hoping to gain from this? "I'm primarily motivated by curiosity at this point"......so what are your secondary motivations? Not trying to be a jerk...just truly wondering why someone would put in the time and effort to do this.......for no reason other than "I want to see what happens". Seems like a lot of investment....for not much return.... I also want this community in particular to have a more informed opinion when it comes to AI, and that requires exposure to it. I'm happy to take help in that in any way I can. |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
Originally Posted By NvrBrdWes: Ran this up the flag pole, not something we would approve at this time. I really appreciate you offering and checking in first. Mind if I reach out to you directly to explore some related ideas further? |
"The state is not the solution. It is the problem." --Javier Milei
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
"If this is how the state treats its law-abiding citizens, it doesn't deserve to have any"
--Solzhenitsyn
|
Originally Posted By SimonPhoto: I think it would be a useful tool for those that want it, and would likely reduce overall exposure to AI content for those who don't (since they can just blacklist the account). I also want this community in particular to have a more informed opinion when it comes to AI, and that requires exposure to it. I'm happy to take help in that in any way I can. Why do you want this community to have more exposure to AI? Why does it matter to you what gun forum members think about AI? Just seems strange to me... |
| You realize the rest of us are going to abuse it. |
ProTip: Never pick a fight with a tree planter. He swings a hoedad for 9 hours a day, carries 30-40 pounds of seedlings on his back, and has 12-14 friends just like him hiding out on the other side of the junk butt pile. They will fuck your shit up.