A new traffic light system at Cheshire Academy is offering a practical model for integrating artificial intelligence into classrooms without banning it outright or letting it run wild. The private Connecticut school has developed a color-coded approach that tells students when AI tools like ChatGPT are allowed, limited or prohibited on assignments. This approach, detailed in Making AI Work, a limited-run newsletter from MIT Technology Review, highlights the growing need for clear policies as generative AI reshapes education.

What You Need to Know

The rapid adoption of generative AI caught many educators off guard, leaving them without clear policies. Organizations from OpenAI to UNESCO have encouraged AI use, but teachers often feel confused about implementation. Schools like Cheshire Academy are pioneering methods that emphasize reflection and transparency rather than blanket bans. These experiments could shape how AI is taught and used in education nationwide.

The Traffic Light Framework

At Cheshire Academy, assignments now carry one of three labels that signal how students may use AI. Green means full AI use is allowed. Yellow lets teachers permit some tools while banning others, such as allowing spell-check but not a chatbot. Red bans all AI use. This system forces students to think critically about when AI is appropriate, a skill that advocates say is essential for the modern workplace.

  • Green: AI tools like ChatGPT or MagicSchool may be used freely for the assignment.
  • Yellow: Some AI tools are permitted; others are prohibited, with specific guidance from the teacher.
  • Red: No AI use is allowed; all work must be entirely the student's own.

Teacher Tools and Resistance

Many instructors at Cheshire Academy use generative AI to prepare class materials. The school has trained staff on general prompting techniques and the limits of large language models (LLMs), including the risk of incorrect or biased outputs. Popular tools include general-purpose chatbots like ChatGPT and specialized platforms like MagicSchool, which offers a rubric generator and lesson planning features. For some teachers, however, the technology still doesn't fit their workflow. Miriam Przybyla-Baum, who teaches French, has built decades of classroom materials and sees little need for AI in her own preparation. She instead focuses on teaching students to evaluate AI's output critically. Some educators share Baum's caution, worried about privacy, personalization and the loss of student voice.

Why This Matters

The Cheshire Academy experiment matters because it provides a replicable template for schools facing the same dilemma: how to acknowledge AI's existence without letting it undermine learning. The traffic light approach gives teachers agency while forcing students to reflect on their own use of AI. This is especially important as organizations such as OpenAI launch student-specific versions of ChatGPT and as companies like MagicSchool expand their education offerings. Without clear policies, schools risk either banning AI entirely and missing its potential or allowing unchecked use that erodes academic integrity. The model from Cheshire Academy suggests a middle path that emphasizes AI literacy over fear or uncritical adoption.

Broader Implications for Education

For educators, the key takeaway is that AI policies need to be granular and adaptable. The traffic light system works because it allows nuance. As more schools experiment with similar frameworks, the definition of "smart AI use" will continue to evolve. The Cheshire Academy case study also underscores the importance of teacher training. Without proper support, even the best system will fail. For now, the school's approach offers a concrete example of how to encourage smarter AI use in the classroom while preserving the human elements of teaching and learning.