That's my same perspective with large projects, actually. And I explicitly tell them they can use AI for their final projects as long as the end result is good. I do this because the large projects are complex enough that they aren't gonna be able to use AI to get from concept to final product.

But what if it's a short essay question, where I explicitly tell them to stick to 3-4 sentences, and it's worth like 0.1% of your grade? I feel like someone who uses AI in that case is just lazy. The exercise is just to get you to think for yourself for a few minutes and they won't even attempt it despite low stakes.