About AI

The push to introduce generative AI into classrooms is moving faster than the evidence and the safeguards. AI must earn its place in our classrooms through independent research before its widely adopted. Until this technology has been proven safe, effective, and appropriate for young learners, the precautionary principle must guide school policy.

The research isn't there yet.

There is currently no conclusive, high-quality research demonstrating that generative AI improves student outcomes. What is emerging, however, is cause for concern: a growing body of evidence suggests that reliance on these tools may impair cognitive development and critical thinking skills in both students and adults. Introducing unproven technology into learning environments before we understand its long-term effects is an experiment we shouldn't be running on developing minds.

AI tools make significant errors, and students can't catch them.

Large Language Models (LLMs) have known error and hallucination rates of 30% or more in certain contexts. Even platforms marketed directly to schools, like MagicSchoolAI and SchoolAI, acknowledge that their products will make mistakes. For students who are still building foundational knowledge, these inaccuracies can easily go unnoticed and quietly undermine their understanding. Students are not yet equipped to evaluate the validity of AI-generated content. Exposing them to confidently stated misinformation at a critical stage of learning carries real academic risk.

Student safety cannot be guaranteed.

Generative AI remains largely unregulated, and there is no guarantee these tools are safe for children. Students may be exposed to inappropriate or harmful content and interactions. Even the developers of these tools openly acknowledge these risks. Beyond content concerns, there are growing reports of AI-related mental health impacts, including documented cases of AI-induced psychosis in both adults and children. The long-term psychological effects of early AI exposure are simply not yet known.

Not anti-technology. Pro-student.

We fully support teaching students about artificial intelligence and preparing them to engage with these technologies responsibly and judiciously. But that preparation must begin with a strong foundation in reading, writing, and critical thinking. Students need these skills before they can meaningfully or safely engage with generative AI tools.

Introducing these tools must first begin with critical, structured, age-appropriate digital literacy programs that don't necessarily require access.

Articles

The New Yorker, What Will It Take to Get AI Out of Our Schools?

"The main arguments against the use of generative A.I. in children’s education are threefold. The first is that L.L.M.s encourage cognitive offloading before kids have done much cognitive onloading—that is, if these tools cause atrophy of thought in adults, then we can scarcely overestimate the potential effects on a brain that has not developed those cognitive muscles in the first place."

"The third complaint against the use of A.I. in schools is that it confuses ends and means, privileging the most efficient route to the correct answer, the crispest thesis statement, or the neatest drawing over the messier and less quantifiable process of building a thinking, feeling person. “We are potentially undermining complex thinking, changing the development of sociality, and mistaking the learning goal,” Mary Helen Immordino-Yang, who is a professor of education, psychology, and neuroscience at University of Southern California, told me. “We are cutting off learning at the knees.”

Chalkbeat, Why Sal Khan’s AI revolution hasn’t happened yet, according to Sal Khan

"Kristen DiCerbo, the organization’s chief learning officer, said AI can only respond to students based on what they ask. And it turns out, she said, “Students aren’t great at asking questions well.” DiCerbo was initially hopeful that AI would be able to personalize instruction to students’ needs and interests. That hasn’t happened. “So far I am not seeing the revolution in education,” she said."

Education Week, Real-Time Data Shows Exactly How Students Use AI on School Technology |

An analysis of 1.2 million student AI interactions on school-issued devices in more than 1,300 districts found that 1 in 5 interactions involved "cheating, self-harm, bullying and other problematic behaviors".

The Guardian, Google faces lawsuit after Gemini chatbot allegedly instructed man to kill himself

Google Gemini, the preferred model in K12 schools, allegedly led a 36-year old man with no documented history of mental health issues to suicide. What is striking about this incident is that all the "guardrails" were functioning as designed. The system even assured him guardrails were in place when he grew uncomfortable, and yet Gemini continued to lead him down a dark path.

MLA, Statement on AI and Assessment

“Strongly opposes the use of generative AI technology as a primary means of assessing student writing. At no time should AI be used as the sole means of “grading” student essays”.

Research & White Papers

Ben Williamson et al., National Education Policy Center, Time for a Pause: Without Effective Public Oversight, AI in Schools Will Do More Harm Than Good

Mary Burns et al., Brookings Institute, A new direction for students in an AI world: Prosper, prepare, protect

"Ultimately, the task force found that at this point in its trajectory, the risks of utilizing AI in education overshadow its benefits. This is largely because the risks of AI differ in nature from its benefits—that is, these risks undermine children’s foundational development—and may prevent the benefits from being realized."

Stanford Scale Initiative, Understanding the Evidence Base for AI in K12 Education

Analysis of 800+ papers finding only 20 "high-quality causal studies". "There are no high-quality causal studies of student AI use conducted in U.S. K-12 classrooms. Most studies examine short-term outcomes rather than long-term learning."

Perlis RH, Gunning FM, Uslu AA, et al. Generative AI Use and Depressive Symptoms Among US Adults. JAMA Netw Open. 2026

”...daily or frequent use of AI is significantly associated with greater levels of depressive symptoms. The odds of reporting moderate depression were 30% higher among people who used AI each day.” (pre-print)
Harvard Kennedy School, How is AI use connected with depression?

Grace Liu, et al., AI Assistance Reduces Persistence and Hurts Independent Performance

"In a series of large-scale human experiments, involving arithmetic and reading comprehension, we find that AI assistance improves immediate performance, but it comes at a heavy cognitive cost: after just ∼10 minutes of AI-assisted problem-solving, people who lost access to the AI performed worse and gave up more frequently than those who never used it. These findings raise urgent questions about the cumulative effects of daily AI use on human persistence and reasoning. We caution that if such effects accumulate with sustained AI use, current AI systems — optimized only for short-term helpfulness — risk eroding the very human capabilities they are meant to support."

Nataliya Kosmyna, et al., Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task (pre-print)

"Over four months, LLM users consistently underperformed at neural, linguistic, and behavioral levels. These results raise concerns about the long-term educational implications of LLM reliance and underscore the need for deeper inquiry into AI's role in learning" (pre-print)

Yanhui Wu, et al., The Generative AI Learning Penalty: Evidence from Chinese Secondary Education (pre-print)

A 30-month study of 26,811 Chinese secondary students found: "AI adoption raises homework scores by 18% and reduces completion time by 30%, but lowers monthly exam scores by 20% within six months. High-stakes entrance-exam scores fall by 18 and 24%, with the full penalty emerging only after about two years."

Sina Rismanchian, et al. Faster Completion, Less Learning: Generative AI Reduced Study Time on Math Problems and the Knowledge They Build (pre-print)

Researchers found that after 2023, time spent on word problems (those that are easy to toss into ChatGPT) dropped dramatically in non-proctored assessment (practice in which AI is accessible) and there was a large increase in correct responses. During the same time correct responses to math word problems in proctored settings (no AI access allowed) fell by 25%.
“Generative AI tools now sit adjacent to nearly every educational task a student undertakes. Our findings are strikingly alarming: students are spending substantially less time on AI-susceptible problems, and this shift is associated with a substantial decline in retention of the underlying concepts.”

Mei Tan, et al., Marked Pedagogies: Examining Linguistic Biases in Personalized Automated Writing Feedback

An analysis of 600 eight-grade essays: “Our results reveal systematic, stereotype-aligned shifts in feedback conditioned on presumed student attributes—even when essay content was identical. Feedback for students marked by race, language, or disability often exhibited positive feedback bias and feedback withholding bias—overuse of praise, less substantive critique, and assumptions of limited ability. Across attributes, models tailored not only what content was emphasized but also how writing was judged and how students were addressed.”
EdWeek, AI Changes Its Feedback on Students’ Writing When It Knows Their Race, Gender