Education has always been slow to change. The basic structure of the classroom, a teacher at the front, students in rows, a fixed curriculum delivered on a fixed schedule, has survived every wave of technology that promised to transform it. AI is different. Not because it is more powerful than what came before, but because for the first time there is rigorous evidence that it changes learning outcomes in ways that matter.
That evidence is arriving at a complicated moment. AI is feared to be part of the problem: students offloading thinking to machines, test scores falling, intellectual stamina declining. And yet the research is showing that AI tutoring designed around active engagement rather than passive answers does the opposite. It builds the capacities schools are worried about losing.
A study published by researchers at Harvard and MIT conducted a randomized controlled trial measuring college students’ learning when content was presented through an AI-powered tutor compared with an active learning class. The study enrolled 194 students across two consecutive weeks, studying surface tension in one lesson and fluid flow in the other. Every student experienced both conditions: a 60-minute active learning class with peers and instructional staff, and an at-home lesson delivered through a custom AI tutor called PS2 Pal.
The study found that students learned significantly more in less time when using the AI tutor, with median learning gains more than double those of students in the traditional active learning classroom. They also reported feeling more engaged and more motivated. The control condition was not a passive lecture. It was an active learning classroom, considered best practice in physics education. The AI tutor outperformed best practice, not just the status quo.
The study’s main lesson is that instructional design matters as much as the technology itself. PS2 Pal worked because it was built around verified content, structured practice, timely feedback, and self-paced learning. The researchers say the findings offer empirical evidence for the efficacy of a widely accessible AI-powered pedagogy in significantly enhancing learning outcomes.
Teachers as Designers, Not Bystanders
A Brookings Institution report published by the Center for Universal Education found that AI’s greatest potential in education comes when educators are active co-creators of the tools rather than passive recipients of them. The report argues that successful AI companies meaningfully engage educators and other stakeholders from the earliest stages, producing tools that align with research-based learning practices and earn genuine teacher trust.
We’d love to be your preferred source for news.
Please add us to your preferred sources list so our news, data and interviews show up in your feed. Thanks!
The report highlights examples where AI is already extending student agency in measurable ways. Junior Achievement programs are using AI agents to give secondary school students the opportunity to practice pitching a product to an investor, with real-time feedback before presenting to actual business leaders. Online experiential learning simulations are providing students opportunities to practice real-world skills from job interviews to entrepreneurship at a scale no individual classroom could support alone.
The report frames the most effective AI in education around what it calls agentic teaching: tools that prompt students to think through problems rather than supply answers. AI designed with educators and students from the start, the report argues, produces tools that genuinely advance learning rather than repackage existing approaches in a new interface.
Personalization at the Scale Schools Could Never Achieve Alone
An EdSurge article surveying major K-12 trends for 2026 found that schools are deploying AI across a growing range of functions as districts search for solutions to declining academic performance and chronic absenteeism.
Jim Bowler, general manager of classroom solutions for the assessment company Riverside Insights, told EdSurge that personalizing instruction is the Holy Grail of education and that AI is the most promising tool available for delivering it at scale. Schools are mining student data for insights that were not possible before, with attendance tracking serving as an early warning system that can alert schools before a student begins to fall behind rather than after.
The districts moving fastest are those treating AI as one tool among many, combining it with science-backed lesson plans and direct student support. The opportunity is in using AI to personalize at the scale that individual teachers cannot achieve alone, freeing classroom time for the collaboration, mentorship, and deeper thinking that no AI tutor can replicate on its own.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
Add as Preferred Source
