Schools are experimenting with AI with little evidence or policy to guide them

Schools across the U.S. are rapidly experimenting with AI in classrooms, deploying chatbots, humanoid robots and teacher-facing tools despite limited evidence and scant policy. In Dayton, Ohio, an alternative high school paid SchoolAI about $72,600 for multiple AI tools including a social-emotional chatbot named Jordan, designed with consultants to elicit students’ feelings by asking for advice and grading responses; students reported building relationships with the bot, and administrators said chat logs were accessible to school staff, an external evaluator and company employees. District leaders renewed the contract for roughly $78,650 after citing positive feedback, even as outside evaluators acknowledged that measurable success depended largely on perceptions rather than hard outcomes.

A Stanford review of more than 800 studies found little rigorous research on how AI improves student learning and suggested purpose-built tools show more promise than general chatbots. With state and federal authorities largely hands-off, districts are proceeding unevenly: Wichita is staging cautious pilots - teaching younger students about AI, allowing limited image- and code-generation tools in middle school and piloting an AI math tutor in high school - while Salamanca paused plans to deploy a $58,000 Realbotix humanoid after community outcry over privacy and cultural concerns. Education experts urge clearer metrics, oversight and higher-level guidance to prevent harms, especially for marginalized students.