AI Megathread
-
I’m open about how I use AI. I don’t use it to write my poses for me, but I’ll occasionally ask it for suggestions to punch up a line that isn’t coming out the way I want it to. I’ll also ask it to generate scene ideas, suggestions for background npcs, things like that. It’s been really good for inspiration on those days when I have the time and energy for RP, but I don’t have any ideas popping out at me.
I mostly use it as a filing system so I can cross-reference things to make sure I don’t forget about something that happened in a scene two years ago. It took a while to train it up and make sure it understood game theme, but it really is helpful for my old-person brain.
-
@InkGolem Do you share their writing with it as well, for example sharing the scene in progress?
-
I never share a scene in progress, and I only share ones that are publicly viewable on the internet.
-
@InkGolem I suppose the robots are out there scraping the mines regardless. I’m glad you seem to recognize that as a boundary for people though. I think that would be a particular hot spot for many people in the RP world.
-
@InkGolem said in AI Megathread:
I never share a scene in progress, and I only share ones that are publicly viewable on the internet.
Just because something is available to potentially get scraped automatically doesn’t mean you need to help it out by personally spoon-feeding other people’s writing into the GenAI database.
-
@Tez Absolutely, and I don’t give it logs that I wasn’t involved in, either. It really has been great for “oh fuck, what was that NPC’s name again”, because it keeps an updated running list of both pc and npcs from my logs, and it automatically updates its analysis of their current relationship with my pc. my memory isn’t what it used to be and it saves me from me forgetting that i had, in fact, interacted with another pc in the past. if i can’t recall, i’ll ask it “have i had a scene with x” and if the answer is yes, i’ll ask it for a brief summary of the interaction.
Like I said, I use it as an extension of my brain, filling in the gaps that my own neurons don’t seem to be able to hold onto as well as they used to. I wouldn’t want to RP with someone who was just using a LLM to write their poses, either. I just was pointing out a utility that I’ve found.
When it punches up lines, it pulls on my own writing for inspiration, and yeah, it sometimes hallucinates, so that’s why I, an actual human being, use my discretion about what to take from its suggestions.
I’ll concede to Roz’s point about ethics. That can be a valid point, and it’s one I’ll think about.
-
@Roz There’s also the case where someone might have set up a robots.txt file to reject the robots, but it would still be visible to people. I have done that, for example. You wouldn’t necessarily know if it can be scraped or not.
I’d personally have some real reservations introducing someone else’s writing to the great chatbot in the sky.
-
Good points all around. I’m going to think on it and maybe just give summaries and bulletpoints going forward instead of full logs. Never really thought about it this way and I appreciate the feedback and food for thought.
-
@InkGolem said in AI Megathread:
I never share a scene in progress, and I only share ones that are publicly viewable on the internet.
Publicly viewable doesn’t mean “use for whatever you want” though. By feeding scenes into the GenAI databanks, you’re allowing the written work of your fellow RPers to be leveraged to generate other AI slop and put other writers out of business. It’s feeding the machine. That’s the harm of which I speak.
-
@Faraday Point absolutely taken. This conversation has, in a rare instance that I can’t recall seeing on the internet in many years, changed my understanding and stance on this.
-
@InkGolem I appreciate you having an open mind about it.
ETA: Also this is an example where it MIGHT theoretically be possible to find an ethical LLM to do what you need, with a private model so the data never becomes fodder for the LLM in the sky. The mainstream models like ChatGPT are just not that tool.
-
@Faraday Absolutely. I am always willing to learn and I am willing to admit when I don’t have the expertise to make a good call and adjust accordingly.
-
Whelp, now to add ‘Please don’t feed my scenes into an LLM’ to my +finger, too…
-
Notepad doesn’t need Copilot.
-
@Hobbie I am glad that it has that “you didn’t save this but we’ve kept it in the buffer for you” when files weirdly decide to open in notepad instead of notepad++. Which isn’t AI, but is neat.
-
@Pavel that is a genuine QoL feature I use daily.
But I don’t need to have Copilot/Gemini/whatever advertised to me every time I open something that isn’t a video game.
I don’t even read the AWS roadmaps anymore because it’s endless updates to Q and nothing about bringing their existing functionality out of minimum-viable-product stage.
AI has its effective areas of use, but it is far narrower than what has been marketed and sold.
-
@Hobbie You mean you don’t want me to tell you your post was PoWeReD bY AI
like has been thrown at us for every. single. fucking. thing. that’s on a computer these days?But on the serious, I’m 100% with you. Tired of seeing Gemini, Copilot, and all the others flung at me as if they’re trying to honeypot me with the latest tech buzzwords*. I don’t need AI to do my job for me; I do need AI to do things for me that are far beyond my capacity. Like massive quantities of data analysis and collation.
AI is way oversold and way underdelivered, and I dare say that the “AI” products being thrown out are not, strictly speaking and technically (pun!) actual AI, but more a heuristic algorithm like pathfinding algorithms such as A*.
* ETA - It’s so bad I just had this comparison come to mind: “They’re acting like MU* sex pests pretending not to be sex pests.”
-
@dvoraen This is one of my favourite reads lately: https://ludic.mataroa.blog/blog/i-will-fucking-piledrive-you-if-you-mention-ai-again/
-
@Hobbie I love that one.
Friend quoted me this recently:
"It’s like ChatGPT has read everything on the internet, and kind of vaguely remembers some of it and is willing to make up the rest.”
There are so many documented instances of LLMs making up nonsense. Citing books that don’t exist. Making up fake lawsuit citations. Misrepresenting articles written by journalists. Making up fake biographical details. The code it spits out is often garbage (or, worse, wrong in subtle ways). And that’s not even touching on all the random stupidity where it tells people to use glue in their pizza or incorporate poison into their recipes.
The whole GenAI industry is most likely just a big bubble built on a con.
-
People call this “hallucination”. I think we should stop letting them assign a new name to an existing phenomenon. The LLM is malfunctioning. It is saying things that are wrong. It is failing to do what it was designed to do.
