Prioritizing Student--Led AI Policy Over Administrative Bans

The mock student senate held by the Edward M. Kennedy Institute shows a clear problem: adults are stuck on AI policy, leaving schools to manage without guidance. By having 98 high school students draft national recommendations, the project shows that effective AI policy is not about banning tools, but about setting rules that value human thinking over cognitive offloading. This exercise shows that while school leaders struggle to agree, students have a practical grasp of the risks. For educators, involving students in policy creates standards that actually work because they reflect the reality of the classroom rather than abstract fears.

The Trap of Cognitive Offloading

The students identified a major risk: cognitive offloading. As Campbell McHenry noted, students are using AI to skip the learning process, treating the tool as a brain instead of an assistant. Many schools respond by banning AI, but the student senate suggests this is a losing battle.

"Without that, it's just going to go unhinged. Everyone is going to stop learning and start relying on AI instead of using AI as a tool. They're going to use it as a brain."

-- Campbell McHenry

When schools ban the technology, they create a cycle where students break rules just to finish assignments. The students proposed a different path: focus on literacy. They argued that AI education should start as soon as students get their devices. This accepts that the technology is already in the classroom, leaving the user's competence as the only thing left to manage.

The Hidden Cost of False Positives

The student senate also looked at the problems caused by AI detection tools. When schools use automated detection without an appeal process, they create a system where students are guilty until proven innocent, which destroys trust. David McConlonaris pointed out that students are often punished by algorithms they cannot challenge.

The student-led policy calls for a change: a mandatory appeal process involving at least two adults. This is a case where administrative convenience, like using a tool to flag plagiarism, creates a long-term cost by damaging trust and risking unfair discipline. The students' demand for human review adds friction to the process, but it protects the integrity of the school disciplinary system.

Designing for Safety, Not Just Compliance

The students' approach to AI-assisted mental health support shows they understand system boundaries. Dylan Dornack proposed that if a student uses a chatbot for emotional support, the AI should be programmed to alert school officials rather than trying to provide therapy.

"It'll alert those people and then those people who go and find that student have them come talk to them, provide them with the resources that student clearly needs in that moment."

-- Dylan Dornack

This insight maps the process: the AI is a way to gather signals, not a way to provide solutions. By routing the interaction back to human intervention, the students prevent the system from creating a false sense of security. It recognizes that AI can identify a crisis but cannot resolve it, a distinction many adult-led policies miss.

Key Action Items

• Shift from Detection to Literacy: Stop spending money on automated detection tools that produce high false-positive rates. Instead, put those funds into AI literacy programs that start when students first get their devices. (Immediate)
• Establish Multi-Adult Appeal Boards: If using AI-assisted grading or detection, require a two-person human review for any flagged work before issuing consequences. This protects the school's credibility. (Immediate)
• Implement Human-in-the-Loop Mental Health Protocols: Ensure any AI tool used in schools has triggers to alert human staff when keywords related to self-harm or crisis are detected. (Next 3-6 months)
• Differentiate by Grade Level: Follow the student senate model: ban AI for graded writing and testing in early grades to build foundational skills, while allowing it for editing and study support in older grades. (Next 6-12 months)
• Prioritize Policy over Perfection: Recognize that most districts currently have no policy. Adopt a framework that provides clear guidelines now instead of waiting for a perfect consensus that may never happen. (Immediate)