AI In Poses
-
IIRC @Tez did have some issue with player(s) using AI for stuff over on that there Demon (and others) game they ran. Their input might be warranted here too, if we’re having a sensible conversation about it.
-
@Ashkuri said in AI In Poses:
You ask them if they’ve used AI and they say ‘no that’s my writing.’ Which seems super unlikely, but rare is the confronted player who just says “ya got me.”
What happens next?
You’re right that players rarely admit to using LLMs. Fundamentally in your scenario you have a player who has broken a rule. If someone breaks a rule, I would ban them. There’s not a lot of nuance in that.
The nuance, of course, is in the question on whether or not you can truly accurately determine whether or not someone is using LLM.
I strongly believe that you can. I honestly have found the conversation in the other thread sort of baffling on a fundamental level. I won’t say that it is always obvious – and this is something I will touch on in a moment – but between AI detectors and human intelligence, you can tell.
Have I caught everyone who uses LLM? Maybe not. Am I confident that everyone I’ve caught using LLM truly did? Yes.
@Faraday said in AI In Poses:
@Ashkuri I doubt I would try to enforce such a policy for individual poses, just as I don’t routinely run other peoples’ poses through a plagiarism checker. But speaking hypothetically…
I might not run things through a plagiarism checker, but I literally have seen people steal descriptions from other people and reuse them on other games. (@Roz for example. Someone stole her character desc from Arx and tried to use it on Concordia. As I recall, the player was disciplined. I am not sure if they were banned.)
We can and do punish players for plagiarism in this hobby, so if we treat them as equivalent, then why wouldn’t we punish them?
If their poses are that nonsensical, probably others will too. Feels like kind of a self-limiting problem to me.
I think you fail to understand how far LLM have come. You aren’t going to get absolute nonsense poses. LLMs are producing writing that grows more and more sophisticated. Like it or not, the technology evolves quickly. I don’t think you can dismiss it by saying that it is going to be obvious nonsense.
I’ve been thinking about how noticeable this will be going forward. As the technology grows more sophisticated, it may become more difficult to detect. It may slip past a threshold where I am confident in my ability to detect. I don’t know. I’ve spitballed ideas about how to deal with it in my head. Some of them are so silly that I won’t derail this thread with them.
For now, though, I can tell. I ban for plagiarism. I ban for LLM. Don’t break rules on my games.
No one post about em-dashes in this thread for the love of god. No one is coming for your em-dashes.
-
@Ashkuri said in AI In Poses:
You ask them if they’ve used AI and they say ‘no that’s my writing.’ Which seems super unlikely, but rare is the confronted player who just says “ya got me.”
I came across this periodically on Shattered (which didn’t allow LLM generated poses) and, while I do think dishonesty is a common response, my experience was more nuanced than that. My first step if I thought something had been generated by LLM (and my experience is it’s never a one-pose thing, it’s something that consistently reads ‘huh a little bit off’ or 'something I made a 90% copy of with the right prompt on ChatGPT), was to page them and ask, ‘Your content <with examples> reads like it was generated by an LLM. What’s going on here?’
If it got to this point, and it did rarely because for me it had to be pretty blatant or chronic, I never got a straight-up ‘no.’ What I got was more answers I think minimizing something as a blatant ChatGPT-copy-paste job. ‘I use editing software I guess it rewrote that’ or ‘I used it to make something longer and better.’ The latter, btw, is the one I think was the most honest and it was ABSOLUTELY an untrue description of what they’d submitted. The LLM background they originally sent in had like three thematic errors that ChatGPT appeared to have just randomed into existence. The shorter (for the better) BG they rewrote was fine.
-
One thing to consider, especially for myself is that there’s a noticeable difference in my posing when I’m focused and super into the RP that’s going on and when I’m RPing while I’m in an emotional low. When I’m not feeling mentally great, my poses tend to become repetitive and incredibly short and direct. It goes from a 3 paragraph pose of very descriptive language to “Dave hands Bob the Temporal Annihilator and says ‘Here you go.’”
It’s possible that others have the same sort of shift in their posing due to mental health, and could create a false positive for AI posing.
-
I saw a lot of people going ‘oh no, people might think I’m AI’ on the other thread but no examples of anyone actually getting incorrectly flagged. I think these are strawmen. Have we seen it happen?
she/they
-
@Tez said in AI In Poses:
I saw a lot of people going ‘oh no, people might think I’m AI’ on the other thread but no examples of anyone actually getting incorrectly flagged. I think these are strawmen. Have we seen it happen?
I haven’t seen it directly but I’ve heard an anecdotal story from someone I trust and play with enough to be pretty confident they aren’t an LLM poser. My understanding is someone ask them if one pose was AI based on Em Dash Paranoia.
-
Has ANYONE gotten banned, not suspected, BANNED, for use of LLM in poses/profiles/etc when they HAVEN’T used it?
This is the only thing that concerns me. I’m a FOOL and was tricked by at least 1 AI app that slipped through. Sorry to catzilla for having to RP with this ai person for a week
I recall you lamenting. -
@Tez said in AI In Poses:
I saw a lot of people going ‘oh no, people might think I’m AI’ on the other thread but no examples of anyone actually getting incorrectly flagged. I think these are strawmen. Have we seen it happen?
I put some of my writing for room descs into an AI detector. Almost every one was marked as AI. Then I tried some from a former builder who I know was using AI (but I hadn’t honed my personal detection methods well enough to spot it, and we’ve since removed all of her descs). And it came back as not AI. This was back in early 2024, though, so maybe detection methods have improved in almost 2 years.
Some things I’ve noticed about LLMs is that they can never create entertaining details. And sometimes it’ll generalize by saying there are a bunch of the thing, or that its qualities are very quality. It spends a lot of time saying nothing of value, while giving the appearance of value. It’s so afraid to offend that it never takes risks. Its prose is cowardly, milquetoast, and superficial. It’s simultaneously compositionally professional and media-illiterate. And it passes those traits on to the people who use it.
Game-runner of Silent Heaven, a small-town horror MU.
https://silentheaven.org -
@Jumpscare said in AI In Poses:
@Tez said in AI In Poses:
I saw a lot of people going ‘oh no, people might think I’m AI’ on the other thread but no examples of anyone actually getting incorrectly flagged. I think these are strawmen. Have we seen it happen?
I put some of my writing for room descs into an AI detector. Almost every one was marked as AI. Then I tried some from a former builder who I know was using AI (but I hadn’t honed my personal detection methods well enough to spot it, and we’ve since removed all of her descs). And it came back as not AI. This was back in early 2024, though, so maybe detection methods have improved in almost 2 years.
I think a lot of us can say this about things we’ve written. Sometimes, especially because as a hobby we do often write a lot , and often in the areas these datasets are trained on, the way that we write CAN look sus. Lord knows we do, and have, and did raise this concern in the other thread.
But has anyone actually disciplined you for the things you’ve sincerely written, though? That’s the case I’m actually interested in, not the anxieties people have that they might accidentally get flagged as AI and banned on an off day. I just don’t think that’s happening.
she/they
-
@Jumpscare said in AI In Poses:
I put some of my writing for room descs into an AI detector. Almost every one was marked as AI. Then I tried some from a former builder who I know was using AI (but I hadn’t honed my personal detection methods well enough to spot it, and we’ve since removed all of her descs). And it came back as not AI. This was back in early 2024, though, so maybe detection methods have improved in almost 2 years.
From playing with both the various detectors and LLMs, it’s gonna vary depending on which one you use and how much text you have to work with (I’ve found it a lot wonkier for descs because they’re a short sample size). I have a paid subscription to one that’s been reliable for me in the past but I haven’t dug into its methodology for false positive/negative rates in a while. At the end of the day it’s mostly a gut check so I’m not just going off pure Vibes when I get to the point where I want to ask somebody if they’re using an LLM. I really, really don’t want to interact with LLM as much as possible in my travels through this hobby. If that seems unfair or Luddite-y? IDK, maybe I am those things. But I would straight-up stop playing at all if this stuff came to dominate the MUSH space. And, while I think I’ve gotten a better sense of it over the years, particularly in creative writing, I don’t want to just go off my gut because my gut is inevitably going to go easier on my friends/players I like than randos I meet in the wild. There is stuff I am 90% sure is from ChatGPT that didn’t ping as it from a Non-Vibes source and I just shrugged and said ‘fair enough.’
-
@Third-Eye It’s cool to be a Luddite. They didn’t hate technology in itself, they were professionals in their own right. They were mad that the automation of their labor never made their jobs easier, the profits all just went straight to the top.
-
@labsunlimited Amusingly I’ve noticed an uptick in Luddite praise across the net lately. It’s cool to be a Luddite again.
-
@Tez said in AI In Poses:
That’s the case I’m actually interested in, not the anxieties people have that they might accidentally get flagged as AI and banned on an off day. I just don’t think that’s happening.
Are you limiting your question solely to MUs? In that case, no, I am not aware of anyone getting disciplined falsely for AI use in MUs.
But in the real world? It’s absolutely happening. There’s no reason to believe it won’t happen here too if people start routinely feeding things into AI detectors.
@Tez said in AI In Poses:
We can and do punish players for plagiarism in this hobby, so if we treat them as equivalent, then why wouldn’t we punish them?
I didn’t say I wouldn’t ban someone for plagiarism if I believed they did it, I said I don’t routinely run poses through a plagiarism detector hunting for violations. Also while I may personally feel that GenAI is a plagiarism machine, I do acknowledge that mainstream society doesn’t see something AI-generated as plagiarism. So I do not attribute the same malice to the action.
-
-
@Faraday There is a human that’s in charge of disinviting people to games. If the human uses their human brain to use a tool to confirm if someone is using AI, based on the results AND their intuition, I don’t know, I think that’s fairly solid. If a human decides to just scan everyone and everything and determine to boot them off their game based on those results…
Like. Okay? If I get banned that way, I don’t want to be on that game anyway? If I go to BMD to whine about it I’d blame staffing decisions. The issue is false negatives, not false positives. At least, that has been our experience running games within the last 5 years.
This thread is about AI in poses. I think detectors are appropriate in this particular environment where human writing really, really matters.
I TOTALLY understand the impulse to have clear, hard lines about this kind of thing. I don’t like arbitrary rules AT ALL. I worry about slippery slopes. But we’re here to write human written lines at each other. This is what it’s all about! This is the CORE of our hobby. If we don’t push back against this with EVERY weapon we have, we might end up RPing with robots. Which people have! It doesn’t feel great!
-
No one asked about this but one of the reasons why Luddites are so maligned as just anti-technology morons is because of pre-Marxist class struggle, which was won by the burgeoning industrialists and capitalists of the early Industrial Revolution period. Many of the luddites, as labsunlimited mentioned, were professionals who knew their craft well. Many of them had probably grown up being taught the craft by parents who had apprenticed them to other expert crafters, and had a great deal of generational knowledge about a specific handiwork. The Luddites weren’t protesting technology, they were protesting the development and financing of machines that created cheap, replaceable, and easily manufactured versions of handcrafted things. I think the main contingent were weavers, but I could be misremembering that. On one level, you could view it as bourgeoisie vs. petit bourgeoisie (industrialists/capitalists vs. small business owners), but many artisan crafters of the turn of the 19th century often lost their business due to cheaply manufactured goods and ended up working in those same factories. Or they wound up in poorhouses, I guess. In the US they probably just starved to death because we’ve always been who we are.
Engels talks a little bit about this in The Condition of the Working Class in England, which is a good historical reference regardless of one’s personal opinions on Marxism or socialism, as you can see elements of this same conflict between the advent of new machines to perform traditionally human labor in our current conversations about AI.
-
@Yam said in AI In Poses:
There is a human that’s in charge of disinviting people to games.
I’m not only talking about staff using AI detectors to ban people, I’m also talking about people running each others’ poses through AI detectors.
If your human gut is saying that something is AI generated, that’s one thing. I just don’t trust these AI detector tools. Everything I’ve seen about them from experts tells me that the fundamental way they work is flawed, and I’ve seen enough drama around false-positives that I don’t want anything to do with them. YMMV.
-
I very strongly disagree these detectors should be discounted. Don’t use the free one google suggests and no other checks, obviously, but the hobby needs to be protected from this slop, at whatever costs. It’s an inherently vulnerable hobby, where everyone is agreeing to submit their actual writing to another person in a scene for reaction. For fun!
It’s not just that LLM writing is hollow, bloated, and uninteresting (though it is), it’s also a breach of that fundamental contract to use it. I have always told people if they aren’t in it to have fun AND make fun for their scene partners, they need a new hobby. If you are using LLMs you are not engaged and you certainly aren’t trying to be fun to rp with.
-
@Yam said in AI In Poses:
Has ANYONE gotten banned, not suspected, BANNED, for use of LLM in poses/profiles/etc when they HAVEN’T used it?
This is the only thing that concerns me. I’m a FOOL and was tricked by at least 1 AI app that slipped through. Sorry to catzilla for having to RP with this ai person for a week
I recall you lamenting.At least it made me aware that people are doing copy and paste of AI in their writings/backgrounds/etc.
I can’t recall if their poses were actually AI but everything in their profiles was.
And then looking back at it (before the website went kablooey) I was like, this is so obvious AI how did I not know?
-
@Faraday The tools work OK from what I’ve seen (as has been mentioned on other threads, Turnitin has modules for this). What it does is flag it as possible - the human has to do some leg work. With essays, for example, you might look at a mini viva. M**s have a pretty low bar, the cost of being booted off the game is…you don’t get to play on that game. That’s all.
