The age of AI has come for K–12 education. A September 2025 RAND survey found that 54 percent of K–12 students indicated they used AI for school, up more than 15 percentage points in two years. Perhaps rapid uptake was predictable, but more striking were respondents’ thoughts about AI’s likely effects: 61 percent of parents, 55 percent of high schoolers, and 48 percent of middle schoolers agreed that “greater use of AI will harm students’ critical-thinking skills.” All three groups were more than twice as likely to agree to the likely harm as the 22 percent of school district leaders, suggesting administrators’ AI sensibilities are out of step with families’ trepidation.
As the Manhattan Institute’s Andy Smarick has pointed out, AI’s entry into K–12 schools isn’t inevitable. When faced with new technologies, from nuclear power to genetic engineering, societies can consider the practical and moral tradeoffs and decide how to proceed. As education leaders make decisions about whether and when to bring AI into classrooms, it’s worth surveying some of the arguments.
In the abundant recent articles about AI in education, there are two broad camps: those who, at bottom, worry AI will supplant student work and learning; and those excited by AI’s potential to supplement learning, both improving instruction and preparing students for an AI-filled future.
Chief among supplant siders’ concerns is that AI’s tempting shortcuts will impede student learning. In the Atlantic, high school senior Ashanty Rosario wrote of the downsides in an aptly titled essay: “I’m a High Schooler. AI Is Demolishing My Education.” Rosario laments the lost camaraderie with classmates who used to struggle together through rigorous coursework; now, most offload the hard work to ChatGPT. My colleague Robert Pondiscio has a complementary warning about the “illusion of learning.” My former colleague Andy Smarick is also worried, arguing that AI’s convenience is too tempting for students not to outsource their thinking to it and goes further, calling for a two-year AI moratorium in public education to prevent a rushed and risky rollout.
Closer to the excited side, Michael Horn argues that the workforce of tomorrow is incorporating AI today, so schools need to familiarize students with the technology. If AI at work is inescapable, worriers like Smarick would unwittingly leave students unprepared for the world they are heading in to, as will educators “more focused on cheating and stopping AI usage than on how they can use AI to make education better.” Adaptation in the age of AI is one job schools shouldn’t drag their feet on.
The workforce argument is weak tea for New York Times columnist Jessica Grose, definitely on team worry, who believes “AI Will Destroy Critical Thinking in K-12.” How well can schools prepare students for an AI workforce when no one can anticipate what jobs will look like, or how those jobs will use AI, in ten years? Critical of Trump’s executive order to push AI adoption in schools, Grose argues the threat to students’ critical thinking outweighs the benefits of preparing students for an uncertain future workforce.
Also on team optimism, Carlo Rotella, an English professor at Boston College, thinks educators can make AI work for them. In “I’m a Professor. A.I. Has Changed My Classroom, but Not for the Worse,” Rotella argues the threat of AI-enabled cheating can push educators to implement good, and largely AI-proof, pedagogical practices. Adaptations like blue book exams, frequent low-stakes quizzes, students showing work, and focusing on class participation work for his students and could work in K–12 as well. While acknowledging that these practices are tougher with larger classes or students who are determined not to learn (what about those who are not determined to learn?), he argues AI isn’t the end of traditional academics.
There is also the promise that AI might improve instructional delivery. My colleague John Bailey argued last year that AI tutors can consistently adapt to every student’s needs, and with their surprising emotional intelligence, may motivate and reassure students better than human teachers—just as AI bots have shown better bedside manners than doctors. AI’s individualized attention and detailed feedback could be more than most teachers can manage on their own, and delaying its use could be a waste.
The debate on whether and how to employ AI in schools is all the more pressing in a time when improvement is needed. On December 8th, we will have that debate at AEI. Google’s Shanika Hope and Alex Kotran of the AI Education Project will argue for the motion “Maximizing School Improvement by 2035 Means Integrating AI into Classrooms Today” while Amplify’s Dan Meyer and Jake Tawney of Great Hearts Academies will argue against. Join us to hear both sides of one of the most important issues for the future of American schools.
