Research on AI’s Classroom Benefits Remains Scarce, Researchers Say
Colleges have bought AI tools at scale, but researchers say independent evidence on AI’s classroom benefits is thin and slow to arrive.
Photo by Yan Krukau on Pexels
Colleges and universities are adopting generative AI tools at scale even though independent research on whether those tools improve learning outcomes remains thin, according to reporting published by Inside Higher Ed on September 14, 2026. OpenAI has sold at least 700,000 licenses to dozens of higher education institutions, including the California State University system, Arizona State University and the University of Maine system, while Anthropic and Google have signed separate deals marketing their models as learning aids.
“The evidence base is almost nonexistent,” Justin Reich, a digital media professor and director of the Teaching Systems Lab at the Massachusetts Institute of Technology, told Inside Higher Ed. Reich said building AI education products and testing them through rigorous studies takes years, and that funding for such research is scarce. A randomized controlled trial – the standard researchers consider most reliable for measuring learning outcomes – is not the type of study the U.S. federal government or philanthropic funders typically support for AI-in-education questions, particularly after cuts to the Institute for Education Sciences.
Why AI research lags the pace of deployment
Stacey Alicea, executive director of the Research Partnership for Professional Learning, said the speed of AI model updates undercuts traditional research timelines. “AI models were changing every six months; now they’re changing every one to three months. Baseline data in these LLMs are changing weekly, and traditional research models cannot keep up,” she told Inside Higher Ed. Alicea also noted that most existing studies test individual tools rather than shared features across the roughly 1,000-plus AI products marketed to students, making it difficult to draw conclusions that apply beyond a single product’s classroom trial.
The California State University system illustrates the scale-versus-evidence gap. After signing a $17 million contract with OpenAI in 2025, CSU surveyed close to 100,000 faculty and students about which AI tools they used most and why; most named ChatGPT. The survey measured adoption patterns rather than learning outcomes.
A preliminary review finds methodological flaws in existing studies
Patrick O’Neill, associate professor of life and physical sciences at Ivy Tech Community College in Indiana, published preliminary findings last month reviewing peer-reviewed studies that claim AI is an effective learning tool. According to O’Neill’s preprint, several of those studies rely on flawed methodologies and misapplied, miscalculated or misinterpreted statistics. A separate set of what O’Neill calls “methodologically effective” studies indicates that access to large language models without guidance is counter-productive to building critical thinking skills.
Carly Robinson, director of research at Stanford University’s Systems Change Advancing Learning and Equity initiative, said the limited causal research available supports a measured approach. “There is increasing evidence that AI has the potential to benefit learning, but using it on its own without intentional design or guardrails probably reduces learning through cognitive offloading,” Robinson told Inside Higher Ed. Winss Solutions previously covered a related Stanford SCALE brief that grades AI tutoring models by how much human involvement they retain, which similarly found that fully automated AI tutoring has the weakest evidence base of the models it assessed.
| Metric | Figure |
|---|---|
| OpenAI licenses sold to higher-ed institutions | At least 700,000 |
| CSU-OpenAI contract value (2025) | $17 million |
| CSU faculty and students surveyed on AI tool use | Close to 100,000 |
| AI tools marketed to students, per Alicea | Over 1,000 |
Background
Inside Higher Ed is a Washington, D.C.-based news outlet covering U.S. and global higher education, publishing daily reporting on teaching, technology, policy and institutional finance since 2004. The September 14, 2026 report draws on interviews with researchers at MIT’s Teaching Systems Lab, the Research Partnership for Professional Learning, Stanford’s Systems Change Advancing Learning and Equity initiative, Ithaka S+R and Ivy Tech Community College, alongside CSU’s own 2026 survey of AI tool use. The reporting follows a wave of large-scale institutional AI contracts since 2023, when OpenAI, Anthropic and Google began marketing education-specific versions of their large language models to universities, and comes as some university leaders – including Dartmouth College President Sian Leah Beilock – have publicly called for deeper AI integration in the classroom even as the underlying research base remains, in Reich’s words, “almost nonexistent.”
Sources: Inside Higher Ed; Patrick O’Neill, EdArXiv preprint (OSF); Winss Solutions (prior coverage)
Featured image: photo by Yan Krukau on Pexels (free Pexels license).
Become a Sponsor
Our website is the heart of the mission of WINSS – it’s where we share updates, publish research, highlight community impact, and connect with supporters around the world. To keep this essential platform running, updated, and accessible, we rely on the generosity of you, who believe in our work.
We offer the option to sponsor monthly, or just once choosing the amount of your choice. If you run a company, please contact us via info@winssolutions.org.
I specialize in sustainability education, curriculum co-creation, and early-stage project strategy. At WINSS, I craft articles on sustainability, transformative AI, and related topics. When I’m not writing, you’ll find me chasing the perfect sushi roll, exploring cities around the globe, or unwinding with my dog Puffy — the world’s most loyal sidekick.
