Banning ChatGPT in schools will be hard, so teachers should treat ChatGPT like a calculator: assume students are using it unless they are physically supervised
OpenAI's new chat bot is raising fears of cheating on homework, but its potential as an educational tool outweighs its risks.
Context & Ripple Effects
Days after the New York City Department of Education banned ChatGPT on school devices and networks over cheating fears, and as an OpenAI researcher told reporters the company was still running experiments to figure out how to battle misuse, this NYT column argues the ban is unenforceable: teachers should assume students are using the bot unless they are physically supervised, and design around that assumption — the calculator analogy.
The argument aged into the arc of subsequent coverage: a Stanford survey of 40+ US high schools found 2023 cheating rates did not rise, while OpenAI pivoted from policing to pedagogy with a free K-12 teacher course and one NYC girls' school began using ChatGPT-generated lesson plans to teach students the tool's pitfalls.
First-order effects
- Teachers lose detection-based enforcement: since banning ChatGPT is impractical, homework assigned outside supervised settings can no longer be graded on the assumption it is unaided work.
- OpenAI's misuse-mitigation problem shifts from building detection tools to helping educators redesign assignments, since the burden moves onto classroom practice rather than software.
Second-order effects
- School systems that followed New York City's device-and-network ban face pressure to reverse or soften it, because a policy that assumes no student access outside school walls does not hold up at home.
- Assessment design becomes a product surface for OpenAI: teacher training, prompt libraries, and classroom integrations turn the same chatbot once treated as a cheating threat into a paid-for or platform-locked teaching aid.
Third-order effects
- If the calculator model holds, education splits between supervised in-person assessment (where AI use is controlled) and take-home work redefined as AI-assisted by default — a structural change to what homework is for.
- The pattern points toward governance norms where institutions stop trying to exclude generative AI and instead set disclosure and usage rules inside workflows, the same accommodation now visible in OpenAI's own teacher-training push.
The trend: Education is moving from banning generative AI to assuming its presence and redesigning assessment around it, with OpenAI shifting from misuse-fighting to classroom integration.