Introduction
Karim Nerz first tested a language model for dialogue and was bored by how reasonable everyone sounded. The conflict arrived neatly, explained itself and left before the scene could become dangerous. Theatre cannot survive that much tidiness.
Karim lives in Berlin and works as a theatre maker and AI dramaturge. His route came through independent stages, precarious fees and the search for new forms of storytelling.
Story of the Path into AI
At first Karim used AI for dialogue variants. The results were smooth, but lifeless. The tool became interesting only when he treated it as material to break, not as an author to obey. It revealed repetitions, clichés and the kind of dramatic compromise a human writer might avoid admitting.
Many people in the theatre scene saw AI as a threat to writers, and Karim understood that fear. He had to show that he was not replacing voices but expanding rehearsal processes. His first stage work placed AI text suggestions into live performance, where actors interrupted, distorted and physically contradicted them.
The first rehearsal was awkward. The machine offered tidy solutions; the actors made them worse on purpose. That was when the scene started to breathe.
Current Work
Today Karim develops formats in which ensembles use AI as a sparring partner. In one rehearsal, a model resolved a family conflict too elegantly. The actors improvised against it and found a harsher, truer scene. Karim now asks systems to produce not only dialogue, but weak dialogue that can be exposed.
His productions make AI visible as a cultural question, not just as an effect. He remains strict about authorship and consent. A rehearsal tool must not become a quiet way to harvest performers’ voices.
Personal Advice
“The machine may bring material, but the conflict has to stay human,” Karim says. He advises artists to use AI where it opens friction, not where it files every edge smooth.
Key Facts
Age and place: 35, Berlin.
Background: independent theatre, precarious work, intercultural perspective.
Entry into AI: stage work that live-deconstructed AI text suggestions.
Focus today: performing arts and AI dramaturgy.
Typical tools: language models, improvisation, rights clarification.
Werkstattnotiz
Karim keeps one scene titled “too solved.” It is the model’s most elegant failure. Whenever a rehearsal becomes overplanned, he reads it aloud and asks the actors where the mess went. The answer usually points to the scene that should be written next.