On an essay by Daniel M. Feige and Arnd Pollmann, and a question Alfred Adler would have asked.
Joseph Kosuth, question. Installation view, “The exchange value of language has fallen to zero”, Casa dei Tre Oci, Venice, until 22 November 2026.
This morning I read an essay in Philosophie Magazin titled “Digitale Entgeisterung”, roughly “digital de-spiriting”. In it, Daniel M. Feige and Arnd Pollmann write about the humanities and what generative AI is doing to them. Their core argument: an AI is not a thinking subject. It cannot even lie, because lying presupposes being able to tell the truth and knowing the difference. The world, the two write, is none of the AI’s business.
Up to this point I nod along. I know there is no one behind the screen who understands me; that is the starting point of all the work currently occupying me. The authors’ second idea also makes sense to me: education is a process that takes place in the struggle to find the words, not a possession you buy ready-made. Anyone who takes the shortcut skips the work of clarifying the thought itself. There is something to that, and I would not want to dispute it.
Then I stumbled.
The two authors describe the student who has AI write his paper as someone who shirks the effort, takes the shortcut, essentially cheats. And education, they say, is a way out into the open that one must first earn by “overcoming laziness and cowardice”. Cowardice. The word sticks with me, because for some time now I have been thinking about its opposite: courage, and how little it can be commanded.
It is not even their word. “Laziness and cowardice” is Kant, from his essay on Enlightenment. The two do not mark it as a quotation, but they deepen the trace: a few lines further on comes “Mündigkeit”, Kant’s word for the maturity we are meant to reach by leaving our immaturity behind. Precisely because it is Kant, the word cannot be dismissed as a slip. To speak of education this way is already to make courage a condition.
Read like this, courage is a demand: be less cowardly, make an effort, and you will amount to something. That sounds reasonable. But it is the language of judgement, and I wonder what that judgement does to the person it is aimed at.
At this point Alfred Adler would probably have asked a different question instead of looking for a flaw of character. Not “Why is this person so lazy?”, a question that already carries the verdict, but “What purpose does this serve? What does he secure for himself by handing the task to the machine?”
Let us imagine the student not as a comfortable cheat but as someone facing an exam he might fail. Perhaps he has learned that his own attempts get graded, compared, found wanting. Perhaps he does not experience the task as an invitation to think at all, but as a test that could expose him: a tendentious apperception that shows him the world in a particular light. The AI then offers a way out that looks safe, a result no one can examine for its weak spots. That is not laziness. It could be a safeguarding tendency, an attempt to seal oneself off against a feared defeat.
Whether that is true in any given case, I do not know. But it is a reading that sees more than the reproach.
Adler never prescribed courage. Kant did. “Sapere aude! Have courage to make use of your own understanding!” The motto of the Enlightenment is an imperative.
And Kant names the cause of immaturity explicitly: it lies “not in lack of understanding, but in lack of resolution and courage”. So he diagnoses a lack of courage and then prescribes “Have courage”. The cure repeats the diagnosis.
Adler sees exactly this futility. Only someone who already has the capacity and merely leaves it idle can follow an imperative. To the discouraged person, though, the one who cannot or believes he cannot, “Have courage” says that all he lacks for maturity is goodwill. It confirms the very verdict he is already cowering under: that as he is, he is not enough. The demand is likely to produce what it denounces: more withdrawal, more avoidance, more reaching for the safe shortcut.
So Adler shifts one word. “Self-incurred” turns hesitation into guilt and locates the cause in the person, in his lack of courage. Adler sees a safeguard there instead of guilt, something the person did not choose as a fault but learned as protection. That is why he answers discouragement with encouragement rather than with a demand. And encouragement is a relationship, not an imperative. You cannot pronounce it like an order; you can only create it.
This turns the instrumental rationality that Feige and Pollmann hold against AI back on their own diagnosis. They do not go wrong by naming a deficit; they go wrong by treating courage as something at our disposal, something that can be produced on demand, like a performance. That is the very logic they hold against the machine.