Add mandatory AI disclosure to PR template

[skyace65]

This was partially taken from PR #118624 and then modified.

Current enforcement of our AI policy (which you can read here) has been tricky due to a number of factors. One of those factors being that while we do have the policy in place, there's nothing at the point of PR creation that explicitly informs users they need to disclose AI usage. We have a link to our checklist but I wouldn't consider that explicit. This makes differentiating between good faith users who aren't aware of the policy, and bad faith users/AI agents willfully ignoring the policy difficult.

By adding a disclosure requirement with zero tolerance for ignoring the question, this makes the policy more visible, and makes deciding what to do when the AI policy is ignored easier. Obviously there's no 100% accurate way to figure out if something is AI or not, but if we come across something with clear and convincing evidence it's AI, that also says in the disclosure that they didn't use AI at all, it's now easier to decide if the account should be blocked or not, as we don't have to worry if that would be too harsh or not.

[allenwp]

In #118855 we settled on having an explicit requirement stated in the template’s comments, rather than having a heading for it. Maybe the more heavy-handed approach of this PR could be used if the comment based approach of the other PR doesn’t work well in practice.

One other thing that was discussed in #118855 was the need to explicitly ask for how AI was used, in addition to requiring it be disclosed.

[skyace65]

nt stated in the template’s comments, rather than having a heading for it. Maybe the more heavy-handed approach of this PR could be used if the comment based approach of the other PR doesn’t work well in practice.

I do not think these changes are worth doing without a heavy handed approach, and I'd apply that to the other things your PR lists as well, like what problems are and aren't solved. If someone isn't going to take the time to read the instructions we give them it should be an automatic close. If they're not going to read something that we put right in front of them, why should we be spending time reading and reviewing their PR?

I also don't see a downside to doing this approach in general.

[Ivorforce]

The main downside to a heavy-handed approach in the template is the risk of alienating contributors. Those not using AI might not appreciate writing "No" every time, potentially interpreting this as an unprompted gesture of distrust or virtue signaling. Some might see it as a reminder or invitation to use AI.

I'm voting to go with Allen's #118855 first, and if it doesn't work out or the AI situation escalates, we can still reconsider this one.