In the "age of AI", I'm not sure we'll ditch the mini projects. I think testing a candidate's ability to _use_ AI to generate code is an important skill. However, it's so new, I'm not sure what the best practices might be.
I'd be curious if anyone has any advice on this front...