Here's the AI clause from the website (https://mitmh2027.com/):

We strongly recommend that you do not use LLMs or generative AI to solve significant portions of puzzles in this hunt. These puzzles were written and tested by humans and are designed to be solved by humans. We believe that you will have a more fun and rewarding experience if you refrain from using AI to bypass parts of the hunt.

We specifically urge you not to paste puzzle content directly into LLM chat interfaces (This recommendation does not apply to using coding agents to write code for puzzles that require programming). Additionally, we will reject submission task content that is generated with AI.

Like, it's not an outright ban (besides for submission tasks) but that's probably due to recognizing that such a ban would be too hard to enforce

since it's a reccomendation rather than a ban I'm inclined to resolve no. Open to being convinced otherwise though.

I guess I mostly read this recommendation as mostly stopping LLM use on puzzles for teams that would obey a strict prohibition anyway, even if it stops short from a full ban

This is tough. I think it’s clear that using LLMs to destroy the puzzles is against the spirit. It’s not technically a ban, but it kind of feels like one.

I feel very uncertain how I would resolve this if I were the judge.

@phenomist The carve out for programming suggests that they expect that there are some uses where LLMs are the correct tool for the job. Additionally, it seems like uses like “Please define word X” or list some rhymes for X” or “Show me the Braille alphabet” would all be fine. If you could do it with another tool instead I don’t think they are discouraging that. So it’s a limitation, but I wouldn’t call this a ban

@JimHays I mean since it was actually asked in advance, the GPH 2025 AI rules:
> You may not ask an LLM or other generative AI to solve a puzzle or a significant part of one. Treat this like our previous rule about not looking at source code: it bypasses our intended solving experience and goes against the spirit of the hunt. More limited uses, such as doing research into a topic you’re unfamiliar with, or interpreting an individual crossword clue you’re stuck on, are left to your discretion.

This, which would be an unambiguous YES, has carveouts for things like topical research or individual crossword clues (it doesn't have a programming carveout because there's less of an expectation to have a programmer for a team in GPH than in MH, so puzzles are written with that in mind)

@phenomist Ah good to know. I didn’t know the GPH one had carve outs as well

I don't believe that this will be a particularly likely occurrence, but I consider Providence to be somewhat more likely as a team to prohibit AI than Cardinality, even keeping capabilities equal

What if only some LLMs are banned, or they're banned only for specific uses?

@derikk I added some hypotheticals in the description. Feel free to present more hypotheticals and I will answer them. I don't particularly believe in trying to spell out every possibility in advance as I've found that it's a little too restrictive / unimaginative about the future. I'm going more for a "spirit of the question" resolution.