I think it would be good to. I generally think of AI as being a cognitive science, whose ultimate goals don't fit that closely with TCS, but the practice of TCS is usually much closer to here than the other cognitive sciences, and I think AI questions are probably a good fit for here.
Examples of questions that would be good here:
- Knowledge representation;
- Theory of neural nets;
- Concept analysis;
- First-order learnability problems.
Examples of more problematic questions:
- Anything philosophical, e.g., about characterising what is an AI;
- Image recognition;
- Natural language processing.