Cheating leetcode interviews with AI doesn’t seem that innovative to me, just adding dishonesty to a broken practice. Destruction is always easier than creation.

Also, as someone who frequently designs and runs SW interviews, it’s totally possible to run interviews that test actually important SW skills like OO design, error handling, and using APIs, which AIs still fail handily.

If you want to do something cool, make an AI to refactor your codebase for maintainability and security.