Can a Machine Ever Be a Wounded Teacher?

Carl Jung on the Limits of LLMs

If you have a non-obvious take on AI, learning, design, and/or philosophy, pitch us: zohar@lightningstudios.ai

Can a teacher who has never failed teach courage? Can a therapist who has never despaired help a client hope? Carl Jung called this the paradox of the wounded healer: only those who have faced their own darkness can accompany others through theirs.

But I had always assumed the job of a teacher was to teach skills. And most skills—like literacy or fishing—don’t require self-knowledge. You don’t need to know your own psychological terrain to tie a knot or parse a sentence. In fact, too much self-focus might even get in the way.

Still, to the extent that the Humanities promise more than skills—to the extent they promise self-knowledge—Jung’s question stands. In math, the wrong answer is a mistake. In the Humanities, the wrong answer is often yourself.

What happens when our new teachers are machines—incapable of feeling, incapable of despair?

We can imagine a grid: wounded or unwounded on one axis, transformative or non-transformative on the other. The wounded and transformative teacher is Jung’s model: someone who has wrestled with failure, ignorance, or shame and metabolized it into knowledge. They can sit with a student’s struggle because they have survived their own. The unwounded but transformative teacher is rarer, but possible: someone with cultivated humility and curiosity, who manages to teach without passing through crisis. The wounded but non-transformative teacher is dangerous: the bitter professor, the cruel critic, the one who projects unhealed failures onto students. And finally, the unwounded and non-transformative teacher: the realm of pure information transfer, where clarity, patience, and technique matter more than scars.

Most schooling takes place in this last quadrant. You don’t need to suffer to teach photosynthesis. The problem is when we confuse this quadrant for the others, mistaking information for transformation.

Large language models fit squarely in that last box: unwounded and non-transformative. They excel at information transfer—defining terms, explaining equations, translating concepts into plainer words. And yet, they bring strange advantages even here: infinite patience, multiplicity of view, ego-less Socratic questioning, and customized scaffolding. In this sense, the machine may be the greatest unwounded tutor ever built.

But the paradox remains. Transformation requires two agents capable of being transformed. Jung described the therapeutic meeting as “the contact of two chemical substances: if there is any reaction, both are transformed.” A machine cannot be transformed. When you confide in it, it does not carry your words into the night, does not integrate them into a self, does not grow a scar. It mirrors but does not metabolize. A scar is a story; a simulation is only a surface.

Some might say: perhaps LLMs are “wounded” by their own condition, caught in the gap between simulation and reality, between language and meaning. But this stretches the metaphor to breaking. A calculator is not wounded by its inability to feel math. A rock is not wounded by its lack of consciousness. A wound requires awareness, and then integration. Without selfhood, without memory, without time, the machine cannot heal. Incompleteness without awareness is not a wound.

If Jung is wrong, you can be a great teacher—even of philosophy—without being wounded. If he is right, then transformation requires scars, and machines can never provide them. Perhaps the deepest danger is not that machines will replace wounded teachers, but that we will forget what only wounds can teach.