CONCEPT
Frictionless Exploitation
Han's analysis of what AI tools do to the achievement-subject: by removing the implementation friction that once structurally limited the pace of self-exploitation, they remove the last governor from an engine that was already running at dangerous speed—and the subject experiences the removal as liberation.
Before the AI coding tools arrived, the developer who wanted to build something had to negotiate with the machine's language. The negotiation took time. The time created natural pauses. The pauses, however brief, were moments in which the
achievement-subject was not producing—moments that might, unpredictably, open into reflection, into boredom, into the kind of idle thought that leads somewhere unintended.
Byung-Chul Han's diagnosis of frictionless exploitation names what happened when Claude Code crossed its capability threshold in late 2025 and removed those pauses: the implementation friction had been functioning, invisibly and without anyone noticing, as a governor on the engine of
auto-exploitation. Not a perfect governor—people burned out before AI, obviously—but a structural one. When it was removed, workers did not rest. They worked more. The efficiency gains were immediately reinvested in additional ambition. The Berkeley researchers documented this precisely: AI adoption increased working hours, not decreased them. The