Everyone’s Asking the Wrong Questions About AI and Education

# Summary

This piece argues that the public debate about AI in schools asks the wrong questions. Rather than testing whether AI actually produces lasting learning and skill acquisition, much of the conversation accepts short-term assisted performance as success. The author critiques specific studies, points to real-world anecdotes about student dependency on AI, and raises equity and motivation concerns.

# Problems with the research

One prominent critique is how some studies measure outcomes. A study cited as evidence that AI helps children used AI to give hints during math problem solving and found those students performed better than students who received human hints. The author questions whether completing tasks with AI help is a valid measure of learning. If students can only solve problems with a "crutch," they may not retain skills or perform when the AI is unavailable.

The article also warns about study design pitfalls: small samples, many measured outcomes, and the risk of false positives. A referenced example describes how researchers manufactured a headline result by testing many variables in a small trial, illustrating how easy it is to produce apparently significant findings that don't reflect real effects.

# Access and inequality

The debate often assumes students have the means to use AI: devices, reliable internet, and competent teacher support. The author highlights that disadvantaged students may lack these basics, so introducing AI can widen existing gaps. Without equitable access and sufficient teacher training, AI in classrooms can benefit those already advantaged while leaving others further behind.

# Effects on learning and motivation

Evidence and accounts cited in the article indicate harm as well as benefit. Multiple studies reportedly show students who score well with AI assistance struggle when the assistance is removed. The article cites reports that AI is demotivating: some students refuse to study, arguing AI will give the answer. A clinician's account about medical students describes cohorts unable to read and synthesize papers and relying on phone-based AI to supply answers.

That anecdote includes a detail about students graduating with extraordinarily high debt—$690,000 in one account—underscoring broader systemic concerns about credentialing without competence.

# Examples and cultural cues

# Practical implications for schools and policymakers

# Conclusion

AI can produce quicker or higher scores under assisted conditions, but that is not the same as lasting learning. The author calls for skepticism about headline-grabbing studies, attention to unequal access, and policies that prioritize skill acquisition over short-term performance boosts.