Christian Scherling: When Intelligence Becomes Cheap, What Should Remain Difficult?

When intelligence becomes cheap, what should remain difficult?

Christian Scherling, member of the AI|APE Research School and founder of cerridan | design, does not ask whether students should use AI. He asks which cognitive work we may hand over, and which we have to keep doing because that is how thinking is learned.

His starting point is an economic one: when people of very different ability produce increasingly similar output, the output says less and less about the ability behind it. Economist Jason Potts calls this signal collapse in The University after AI. The MIT report on AI in teaching and learning arrives at the same point from the other side: assessment that only grades the artifact will soon measure nothing.

What follows is not a question of technology but of teaching. Which difficulties are worth keeping? And where is the shortcut a genuine gain?

The full article appeared on 22 September 2026, revised on 7 October, at cerridan.com, where an audio version is available as well.