AI and Student Motivation in Schools: Why Design Matters

AI and student motivation in schools: why design matters

Only 26% of 10th graders say they love school. Meanwhile, 65% of their parents believe they do. That gap, documented before the rise in AI-assisted homework use that RAND has tracked over the past year, is the real starting point for any conversation about AI and student motivation in schools (Brookings, 2025).

Chronic absenteeism tells a similar story. It remains nearly double pre-pandemic levels, with about one in four students missing more than 10% of school days a year (Brookings, 2025). At the same time, rising numbers of middle school, high school, and college students report using AI for homework, and most of those users say they worry about its effects, according to a RAND survey published earlier this year. Many also report uncertainty about what AI use is allowed at school, and the same survey tracked growing numbers of students who believe AI use can harm critical thinking.

What AI and student motivation in schools have to do with assignment design

Three words get used loosely in this debate: motivation, engagement, and learning. They are not interchangeable. Motivation is a student's drive to engage with schoolwork. Engagement covers belonging, interest, and participation. Learning is measurable cognitive gain. The research below speaks to all three, but conflating them overstates what any single study actually shows.

AI didn't create schools' motivation problem. It raises the cost of ignoring it. The evidence doesn't support choosing between better assignments and stronger AI guardrails; it supports doing both, and the rest of this piece explains why, addresses the strongest objection to that view, and lays out what teachers, parents, districts, and students should each do next.

The engagement gap existed before AI, and that limits what the data can prove

In a nationally representative survey of more than 65,000 students in grades 3 through 12, only 29% of 10th graders said they got to learn things that interested them, versus 71% of parents who assumed they did (Brookings, 2025). Only 42% said they used thinking skills rather than mostly memorizing, compared with 78% of parents who believed that was happening. And only 39% reported feeling they belonged at school most of the time, versus 62% of parents.

That data establishes something specific: disengagement predates widespread generative AI use in schools. It does not show that routine, answer-focused assignments caused that disengagement, and it doesn't show AI is exposing a problem that wasn't there before. Treating student-parent perception gaps as proof of an AI impact on student engagement would overreach the evidence.

Still, the gap is worth taking seriously on its own terms. If fewer than a third of 10th graders feel their assignments connect to their own interests, then whatever comes next, whether that's an AI policy update or a new grading rubric, has to start from that baseline, not from what parents assume is happening in class.

Where general-purpose AI can replace the thinking that builds learning

This is where the AI-specific evidence gets more concrete, and more useful for anyone designing assignments. Giving students access to a general AI tool improved homework performance but did not improve exam scores (IES, August 2026). Students who used general-purpose chatbots like ChatGPT or Claude while studying were more likely to show lower exam performance, shallower learning processes, reduced brain activity, and weaker recall, even when they personally found the tools helpful.

A separate comparison found students using a traditional search engine for a research paper produced higher-quality reasoning and argumentation than students using a general-purpose AI chatbot for the same task. None of this is a motivation finding. It's a finding about AI reducing student cognitive effort during the exact tasks that are supposed to build understanding.

The connection is worth drawing carefully: if an assignment only asks students to produce a finished answer, and a chatbot can produce that answer with less effort than the student would otherwise spend, the incentive to sit with the material and work through it gets weaker. That's a reasoned instructional-risk argument, not a proven cause of disengagement. IES names avoiding this substitution, protecting what it calls "productive struggle," as a specific guardrail for anyone designing AI policy or classroom tasks (IES, August 2026).

AI in K-12 education can support learning, with real conditions attached

The strongest objection to tightening AI rules is that AI clearly helps some students learn, and that objection deserves a real answer rather than a dismissal. Students supported by AI-assisted tutoring performed as well as or better than students working with human tutors alone, in research IES cites (IES, August 2026).

But the evidence base behind that finding is thinner than it sounds. A review conducted for IES found only 20 rigorous causal studies of AI's impacts across education generally, with most AI research concentrated in postsecondary settings and causal studies more common in high school than in middle or elementary school (IES, August 2026). A separate synthesis citing more than 80 causal studies found positive effects on student engagement and learning behaviors, but that broader base spans different tools, subjects, and grade levels, and shouldn't be read as confirming K-12 generative-AI outcomes specifically (IES, February 2026).

Design matters as much as access. Hint-based tutoring chatbots that avoid giving away answers performed no better than traditional study methods on exams, meaning a well-intentioned tool still has to be built and used correctly to make a difference (IES, August 2026). Researchers also point to a real gap: there's little evidence yet on whether AI benefits students who lack private tutoring or extracurricular support, which happens to be the group any new tool should be tested against before it's adopted. The same research recommends vetting any tool for privacy and access before adoption, alongside sustained investment in teacher training.

None of this splits neatly into "AI helps" or "AI hurts." It comes down to whether a tool is purposefully designed, paired with teacher expertise, and vetted for equity and privacy before it reaches a classroom. That's an argument for guardrails, not an argument against using AI at all.

What this means for teachers, families, districts, and students

The instructional response isn't complicated to describe, even if it takes real work to carry out. Each group has a different, specific next step.

• Teachers and tutors: Redesign one answer-focused assignment into a process-focused one. Instead of a single take-home essay, try a shorter draft, an annotated comparison of two sources, and a two-minute verbal explanation of the student's own claim. This makes thinking visible; it isn't a guaranteed anti-cheating measure, and any AI-use or disclosure expectations attached to it should match the school or district's current policy rather than a personal rule (IES, August 2026).
• Parents and families: The way parents and caregivers interact with a student at home was found to be twice as predictive of that student's interest and learning as socioeconomic status, and students may perceive AI-generated feedback as less caring than feedback from a teacher (Brookings, 2025; IES, August 2026). A concrete step: ask your child's teacher what a typical assignment requires them to think through, not just complete.
• Districts and schools: Students themselves report ambiguity about what AI use is allowed at school, which is a direct argument for clear, current local guidance rather than leaving the decision to individual classrooms, according to RAND's survey of the American Youth Panel published earlier this year. With rigorous causal evidence still limited, policy should require vetting tools for privacy and access before purchase, not after (IES, August 2026).
• Students: Follow your teacher's or school's AI-use rules as written, and ask directly what needs to be disclosed if you're unsure. Use AI tools to question your own reasoning, such as asking one to challenge a draft argument, rather than to produce a submit-ready answer.

The path forward for generative AI and student learning

The perception gap between students and parents, most 10th graders not loving school while most parents assume they do, predates generative AI, and it won't close through banning AI or embracing it uncritically (Brookings, 2025). The AI-specific evidence points somewhere more useful: general-purpose tools that do the reasoning for students tend to reduce learning, while tools built to scaffold thinking, such as tutoring, hints, and teacher diagnostics, tend to help, though the current review of that evidence is still thin (IES, August 2026).

Students report real uncertainty about how to use AI responsibly at school, based on RAND's survey findings from earlier this year. That's a reason for districts to provide clear local guidance rather than leaving every teacher to write rules alone.

Pick one upcoming assignment or one school AI policy and check it against three questions. Does it require visible thinking, not just a finished answer? Does the current school or district policy say whether AI use is allowed and how it must be disclosed? Has the student asked the teacher directly what's expected? Check your teacher's assignment directions and your school or district's current AI guidance, if one is published, since that guidance may have changed for the current school year.