Sometimes it's good to give them a repetitive project where you give them one task and have them repeat it. I once trained an intern to do molecular biology from basically nothing, he made 50 mutants and tested 25 in three months. He got a publication out of that.

On the other hand, I have an intern now, who I've taught go (she made a cli and learned tdd) and had her set up a tensorflow deep learning pipeline, which is not repetitive and very exploratory.