From dismissal bell to bedtime, most high school students have access to AI technology. Early research and common sense both suggest that offloading too much mental work to AI can weaken the very skills school exists to build. AI technologies also have tremendous potential to help students, teachers, and parents. If used correctly, they can help teachers and parents to support students at home. To maximize potential benefits and limit potential harm, our school district needs an AI policy to guide student use.
I am running for the Tamalpais Union High School District school board. I am concerned that, without guidance and collective action, our students will surrender part of their ability to think. I am also concerned that students are not obtaining the benefits that AI technology offers. Missing out on benefits while putting students at risk would be inexcusable.
Our school district currently has almost no AI policy. In January 2026, an AI task force presented an interim report to the board. The task force was disbanded; no recommendations were made, no significant policy changes were adopted.
If I am elected, I will ask the board to vote with me to direct the formation of a new and different task force, which should act quickly to consult with teachers, students, parents, experts, and administrators and develop a policy that will protect students and prepare them to graduate ready for the future. The task force would have a mandate to move quickly, learn continuously, and provide practical guidelines to assist teachers in assigning work and assessing student performance.
Tam District’s current policy addresses AI only in the context of academic honesty.
The school district currently mentions AI in the “Academic Honesty” section of the board’s policy and the parent/student handbook. Academic dishonesty includes: (1) “Use of Artificial Intelligence (AI) to execute any part of the learning process without approval by the course instructor” and (2) “use of AI to produce all or part of a product submitted in place of the student’s own critical thinking, creativity, writing, or other skill being assessed.”
These definitions of academic dishonesty are vague and unworkable. A student cannot know when he or she is executing “part of the learning process without approval,” unless the instructor has barred all use of AI for any purpose. The second definition is also unworkable; it seems to allow the use of AI to produce content submitted so long as the production does not take the place of the student’s own critical thinking. That is an impossible standard to apply objectively.
One could argue that a subjective, “I know it when I see it” standard is sufficient, but a subjective standard is inappropriate, given how severe the consequences of academic dishonesty are. We are talking about accusing students of dishonesty here. The parent/student guide states, “Students who are suspected of engaging in Academic Dishonesty will be provided Due Process ….” The reference to “Due Process” drives home the major consequences of landing on the wrong side of the AI-use regulations.
Accusing a student of dishonesty may well trigger a time-consuming and bitter dispute, without any clean resolution. Proving that somebody used AI can be difficult and likely will get harder as technology improves.1 Even now, a simple prompt can generate plausible student work. Teachers presumably do not want to spend extra hours serving as linguistic detectives, looking for evidence that students, and not AI tools, generated a submitted insight. Detecting this kind of “academic dishonesty” is a dead end.
An AI policy should focus on mitigating the risk of harm and obtaining the benefits of AI technologies for students.
The urgent reason to enact an AI policy for the Tam District high schools is to preserve students’ critical thinking ability, not to articulate different ways that students may violate an academic honesty code. Researchers are only just beginning to explore the impact of AI use on critical thinking skills. The hypothesis that researchers explore is whether tools like Claude, ChatGPT, and Gemini enable users to offload or outsource cognitive load. The ability to produce student-like output with less cognitive effort creates a dependency, which in turn leads to diminished independent thinking ability over time.2 Researchers sometimes compare the “skill decay” effects of ongoing cognitive offloading to muscle atrophy from disuse.3
A close study of the latest research would be an important part of any task force work. At the same time, when the threat is significant, a policymaking board need not wait until researchers deliver overpowering consensus. We must make our best judgment based on the data we have and the dependency we are already seeing. The problem is not, as some would have it, that AI tools are sometimes wrong, but that they are often right.4 They often provide analysis that is far better than what a beginner could provide. And most students are, almost by definition, beginners. The temptation to “learn” with the assistance of AI tools is overpowering and may be disabling.
While the risk of jeopardizing the development of critical thinking skills is the reason for urgency, any discussion of AI also must acknowledge the potential benefits and the opportunity cost of not using AI. In my profession and as a parent, I have seen how AI can facilitate brainstorming, help review concepts, and carry out research tasks. AI tools can be effective at making education available—if students know how to ask, or perhaps more likely, if tools are configured to serve as a tutor instead of a homework-finishing machine. AI gives students an opportunity to ask unlimited questions. For a student who asks enough questions, it can unlock subject areas that teachers may not have time to address.
AI can help teachers with generating practice sets, giving students faster first-round feedback, and working through the individualized-education and compliance paperwork. It can help parents to work with their children. I personally have found that when I don’t remember a science or math concept, I can get up to speed with AI and then turn back to helping with a problem.
I see AI technology as having massive risks and benefits because it is powerful, like fire and nuclear fission. The key for our school district will be to develop guidelines for proper use, so that the benefits outweigh the costs.
The AI task force should investigate guidelines for evaluation that do not depend on regulating AI use outside school.
I assume that regulating and detecting AI use outside of school is impossible. Therefore, I would ask the board to direct the AI task force to investigate guidelines for teaching and evaluation in a world in which students have ungovernable access to AI outside of school. As some educators argue, this may mean that students should be asked to generate work products that AI cannot produce, whether because the knowledge required to create them is hyper-localized and not available online, or because the product is physical, or because the product is a spoken defense of a thesis, delivered in-person, in the classroom.5 To excel on an in-person exam or other in-class work without a device, a student must learn the material.
The task force should also look at the opposite kind of assignment: work where students are expected to use AI and are told so up front. And the AI students use need not be an unrestricted chatbot. Schools can provide tools configured to build skills rather than replace them, such as one that critiques an essay’s structure without generating any replacement sentences, or one that walks a student through a math problem without ever giving away the answer. The task force should explore whether it can provide guidance on tools specifically configured to promote learning.6
I also would ask the AI task force to explore guidelines for teaching and evaluation when the teacher directs the student to use wholly unrestricted AI to generate work products.7 In this scenario, the paradigm shifts from evaluating what the student alone can produce, and the relevant evaluative question becomes, “Is this good work for a student who is using unrestricted AI to research and produce it?” In this scenario, the question is not whether the writing is good—it should be excellent, given that the student is using AI—but whether the student has understood what problem he or she was asked to solve, whether the student addressed the problem, and what choices the student made in explaining the solution. Of course, a student with access to AI tools might ask for AI input on these dimensions as well. The point is to come up with AI-resilient assessment design, where the questions test creativity and critical thinking even when students use unrestricted AI to answer them.
The AI task force must be collaborative and respect teacher expertise, and move quickly to develop guidelines.
Teachers understand how students actually learn and their judgment will be critical in shaping the policy. Their participation and buy-in are also necessary for success. The task force should also draw on the experiences of students who are already living with these tools every day, parents who see the effects at home, outside experts, and the wider community. The purpose of an AI task force will likely be to develop guidelines and guidance, not stringent rules, to help teachers, students, and parents adapt to a world in which AI changes the parameters and possibilities of at-home and in-class evaluation.
We cannot leave the topic unaddressed. Working together, teachers, students, parents, experts, and the board can support great teaching, give students the real benefits these tools offer, and protect critical thinking that education is meant to build.
Some AI detection tools such as Pangram claim an extremely low false-positive rate. See Miryam Naddaf & Richard Van Noorden, AI-detection tools have made huge leaps forward — how good are they? (Aug. 25, 2026), https://www.nature.com/articles/d41586-026-02569-3; Marijke Van Vlasselaer et al., Who wrote this? Evaluating the reliability of AI detection tools in higher education (June 29, 2026), https://link.springer.com/article/10.1007/s40979-026-00226-w. While false positives may be acceptable in the context of evaluating hundreds of academic papers submitted for publication in peer-reviewed journals, they are problematic in the context of accusing high school students of academic dishonesty.
Gerlich, Michael, AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking (January 3, 2025), https://ssrn.com/abstract=5082524.
See Zara Abrams, How AI is reshaping human skills and thinking (July 1, 2026), www.apa.org/monitor/2026/07-08/ai-job-skills-thinking; Brooke N. Macnamara, et al., Does Using Artificial Intelligence Assistance Accelerate Skill Decay and Hinder Skill Development without Performers’ Awareness?” Cognitive Research: Principles and Implications (2024), https://doi.org/10.1186/s41235-024-00572-8.
Hao-Ping (Hank) Lee, Advait Sarkar, Lev Tankelevitch, Ian Drosos, Sean Rintel, Richard Banks, and Nicholas Wilson. 2025. The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects from a Survey of Knowledge Workers. In CHI Conference on Human Factors in Computing Systems (CHI ’25), April 26–May 1, 2025, Yokohama, Japan, https://doi.org/10.1145/3706598.3713778.
See Jocelyn Gecker, The Rise of AI Tools Forces Schools to Reconsider What Counts as Cheating, Associated Press (Sept. 12, 2025), https://www.ksl.com/article/51374038/the-rise-of-ai-tools-forces-schools-to-reconsider-what-counts-as-cheating (profiling Casey Cuny, 2024 California Teacher of the Year, whose students at Valencia High School now do most writing in class under monitored conditions); To Keep AI Out of Her Classroom, This High School English Teacher Went Analog, NPR (2025), https://www.npr.org/transcripts/nx-s1-5631779 (describing a teacher’s shift to fully handwritten, in-class drafting).
Examples include Khanmigo, MagicSchool and SchoolAI.
The task force would need to consider how a student who does not have access to AI at home would complete the assignment. I assume that most students have access to roughly comparable AI tools at home. The task force should consider whether that assumption is correct.
This is such an important topic and one that I have been spending some time thinking about recently. Thanks for raising it so thoughtfully. I am looking forward to when you are our Trustee and can participate in raising the visibility of this topic for our district.