To prevent LLMs from destroying education, the work must happen in class (blainehansen.me)

In a provocative piece published on July 24, 2026, educators are urged to confront the reality that assignments completed outside the classroom are increasingly susceptible to being done by large language models (LLMs) instead of students. The article argues that this situation creates a competitive environment akin to a prisoner's dilemma, leading to widespread cheating among students. To mitigate this crisis, the author advocates for all evaluative work to occur in a classroom setting where access to LLMs is restricted, a shift that positions educators unfairly as custodians of integrity amidst rapid technological change. The commentary emphasizes that while LLMs can offer theoretical enhancements to education, they often detract from the learning process by enabling superficial understanding rather than deep mastery. Techniques like inverted classrooms and mastery-oriented learning, exemplified by platforms like Khan Academy, are presented as promising alternatives that can foster genuine skill acquisition without reliance on AI tools. By embracing these methods, schools have the chance to reclaim educational integrity and make substantial improvements to learning environments, ultimately transforming a perceived crisis into an opportunity for meaningful reform.