MarioWiki talk:Generative AI

Questions

• If you're looking for a word or (turn of) phrase that conveys a highly specific meaning or nuance, does asking a chatbot for suggestions count in any way, shape, or form as "generative" AI? This proves extremely helpful and infinitely less time-consuming than attempting queries one after the other on Google. Speaking of, it's also become kind of an inevitable "choice" seeing as how you're greeted with an AI summarizer when you google something.
• If you want to automate editing large swathes of code that may otherwise take you forever to do manually (such as changing the layout of a table but preserving its contents), is it allowed to ask a chatbot to do it for you by providing a template? Think poor man's PorpleBot (talk). I'm talking about instances where you'd accomplish the same exact result whether you painstakingly work at it or just whip it into shape with a bot.

Both of these are instances where it's virtually impossible to tell if a user used AI or not, but they also veer a bit too close to what the policy prohibits should it ever be confirmed to be the case. With how stringent the policy sounds, it may be helpful to clarify the wiki's position on these instances. -- KOOPA CON CARNE 16:08, December 21, 2025 (EST)

The reason I wrote it to be like this is mostly as a filter to catch people who are lazy with generative AI, in which we did have extremely obvious cases of in the past (that dude was one of the main inspirations behind this policy page in fact). Obviously, if in the final revision of a page there is virtually no difference between human made and made with consultation of AI, it's virtually unenforceable as you said, but I don't want to add clauses to muddy a basically straightforward message: don't be lazy, we will catch you if you're lazy. People who know how to use it use it to augment their work, which in this case I can agree that it's perfectly acceptable use of it. As for editing large swaths of code to make a table, I haven't had much experience with this myself: in the brainstorming process, I did consult other staff members (and I posted this in the Discord channel for feedback) but I did got pushback from @Camwoodstock when I did suggest that it can help you code tables, who may have more experience with genAI's limitations on coding than I do. Xiahou Ba^{(the Nasty Warrior)} 16:20, December 21, 2025 (EST)

For my own thoughts, using AI to assist you in editing the wiki should be fine (such as providing you pointers that you ultimately fix and write), and if you want us to clarify this, we'll do it. The gist of it is that AI, we emphasize it should be used, if at all, for assistance for effort not as a substitute for effort; it's like lane assist in cars to make you drive better, but you still need to practice all the safe habits of driving such as obeying speed limits, checking blindspots, using lane signals, and so on. When it comes to dealing with users who use AI, people who suspect genAI content should assume good faith first and foremost (I anticipate some users accidentally putting in genAI images and this shouldn't be a bannable offense, as it's a mistake; it's why I added edits discussing some tips how to disseminate content you find online), and people should feel safe to disclose if they used genAI to help writing code. For the most part, people should inspect and fix the resulting product first rather than worry if it's created by genAI. As my sister said, a sloppy result is a sloppy result, whenever a human wrote it or a chatbot wrote it. We just want to filter out people who aren't acting in best discretion or good faith with genAI tools. It's me, Mario! (Talk / Stalk) 16:41, December 21, 2025 (EST)