AI, for now, is pretty good with the following:
Using avatars in place of patients for clinical training
Predict the likelihood of death from surgical procedures
Predict HIV progression
Working in a hazardous environment.
Sifting through a large amount of content.
Finding the shortest path to a solution.
Work non-stop 24/7.
Having said that, my impression is that it should be used below the supervision of a human to avoid any disasters due to biases and noise in data.