There was a thread "How to tell an over-confident student they still have a lot to learn?" https://academia.stackexchange.com/a/17833/49.

If a student is smart some problems may seem to them like "I see you know that 2 + 3 is 5, but what's the reasoning?". So indeed, making problems more complex is the only way to go.

And the "reasoning" part is difficult. We never know if something is a true reasoning, something tangentially relevant, or rather something we were trained to say. It works (or: doesn't work for machine learning in a similar way, vide:

"Speaking as a psychologist, I’m flabbergasted by claims that the decisions of algorithms are opaque while the decisions of people are transparent. I’ve spent half my life at it and I still have limited success understanding human decisions. - Jean-François Bonnefon", as quoted in https://p.migdal.pl/2019/07/15/human-machine-learning-motiva....