My interns always got very speculative projects, the kind that were low probability of working out so didnt want to waste (more valuable) time exploring. If they ever found anything useful it was 100% reworked from scratch.

My preference was for both sides to learn at the end of the day.