How to Encourage Smarter AI Use in the Classroom

The sudden release of powerful chatbots has placed a new burden on teachers, who must adapt to technology that can instantly answer homework or generate essays. Many educators feel there is no clear, universal path forward, leading to varied approaches like training on prompt crafting and implementing classroom frameworks. Schools such as Cheshire Academy are trying student-led initiatives and reflective assignments to encourage smarter AI use.

Key takeaways

• Teachers face increased workload from adapting to and detecting AI use, with no universal guidance.
• Schools employ a mix of general-purpose chatbots and specialized platforms like MagicSchool, with training focused on practical prompt skills and AI limits.
• A ‘traffic light’ system categorizes assignments by AI permission: green for allowed, red for banned, and yellow for limited tool use.
• Reflective assignments, such as critiquing AI edits or peer-reviewing AI-assisted work, help students understand AI’s limitations.
• Teacher comfort varies, with many using AI for administrative tasks like lesson plans, but hesitating on student-facing applications due to quality and privacy concerns.

The Educator’s Challenge: Adapting to AI in a Post-Chatbot World

Powerful chatbots arrived suddenly, adding a fresh challenge for teachers already working long hours. They now need to adapt to a technology that can answer homework questions or generate essays instantly. For many, a clear, universal path forward still seems absent. While groups like OpenAI and UNESCO encourage classroom AI use, many teachers remain uncertain about the best way to handle it.

Detection Adds to the Load

This adaptation includes detecting AI use, which increases teacher workload. Instructors can often spot a student using AI, as models make mistakes most people avoid. Some teachers cite simple giveaways in AI-generated text, like an overuse of em dashes.

Cheshire Academy’s Patchwork Approach: Training and Experimentation

Cheshire Academy does not force instructors to use AI, though the school’s technology coordinator states the “vast majority” of instructors use it in some way. Educators employ a mix of general-purpose chatbots like ChatGPT and Perplexity alongside specialized tools such as MagicSchool, an AI-powered platform built for educators.

Training Focused on General Techniques

Following consultant advice, staff training centered on general techniques for using AI. This training covered practical skills like how to craft useful prompts while also stressing the technology’s limits, highlighting its potential for generating incorrect and biased responses.

Teacher Use Cases and Limits

Teachers now commonly use generative AI to prepare class materials, such as lesson plans and grading rubrics. Some have expressed interest in using it to give student feedback, but concerns over quality, personalization, and privacy have stopped them.

The school also experiments with student-led initiatives. It is piloting a “Student AI Council,” where students create media and lead discussions on healthy AI use, pushing peers to think about how AI should and shouldn’t benefit their community.

Practical Classroom Frameworks: The Traffic Light System and Reflective Assignments

Cheshire Academy has adopted a ‘traffic light’ system for categorizing assignments. In this framework, a green label means AI use is fully allowed, a red label bans any AI use entirely, and a yellow label permits specific tools while banning others. For a yellow-light assignment, a teacher might allow students to use a spell-checker but prohibit messaging a chatbot.

Reflective Assignments on AI’s Limits

Educators at the school also design assignments to prompt student reflection on artificial intelligence. Miriam Przybyla-Baum, a French teacher, has developed techniques for this purpose. In one assignment, students have a large language model (LLM) edit their homework. They then must critique the edits, deciding which ones were correct and which ones removed their personal voice.

Another reflective technique involves peer review of AI-assisted work. In this exercise, students anonymously grade each other’s AI-assisted assignments. As part of the grading, they must annotate which parts of the work they believe were AI-generated.

Tool Landscape: Specialized Platforms vs. General-Purpose Chatbots

Educators are managing a patchwork of AI tools. Specialized platforms like MagicSchool offer a bundled set built for teachers, generating quizzes, worksheets, grading rubrics, lesson plans, and administrative reports through prompts made for each educational task.

Cost and Access Models

MagicSchool operates on a freemium model. While it offers free tools, teachers need a paid individual plan of just under $100 per year for unlimited access and complete records within the system.

Varying Comfort Levels and Use Cases

Not every teacher feels comfortable using AI to generate student-facing text, due to doubts about its effectiveness, helpfulness, or accuracy. However, many find value in the administrative efficiency these tools provide. Alongside specialized platforms, many educators use free, general-purpose chatbots from companies like Anthropic, Google, and OpenAI for administrative tasks.