A researcher is leading a national effort for rigorous, transparent edtech—inviting companies to put their AI tools to the test.
INTERVIEW | by Victor Rivero
For all the noise around AI in education, Ola Ozernov-Palchik, Ph.D. is asking the quiet but essential question: does it work? As Founder and Director of the Evidence-Based AI in Learning (EVAL) Industry Collaborative and Associate Director of Research for the AI and Education Initiative at Boston University, she’s rallying developers, educators, and researchers to put AI-based learning tools to the test. Now, she’s inviting companies to join EVAL’s EdTech Research Challenge, offering independent evaluations that separate what actually works from what just sounds good—and helping turn AI in education from promise to progress.
What prompted you to enter this area of edtech, specifically evidence-based edtech?
I have always been a translational researcher and an advocate for children with reading difficulties. When I served as director of the Mind, Brain, and Education master’s program at the Harvard Graduate School of Education, I worked to ensure that every graduate student in our program could read, interpret, and apply scientific findings to educational practice. This was especially important in educational neuroscience, a field often filled with misinformation. Through my research, advocacy, and teaching, I saw how many educational practices were not grounded in evidence and how harmful unproven methods can be for the most vulnerable learners.
‘Through my research, advocacy, and teaching, I saw how many educational practices were not grounded in evidence and how harmful unproven methods can be for the most vulnerable learners.’
As I became more involved in AI and education, I grew concerned about how quickly innovation was outpacing the evidence. Only about ten percent of educational technology tools have been rigorously evaluated, which is deeply concerning given how widely they are used. That realization led me to launch the Evidence-Based AI in Learning (EVAL) Initiative at Boston University. EVAL aims to build an inclusive ecosystem for responsible, evidence-driven innovation in K–12 education, ensuring that technology truly supports both teachers and learners. Our approach is grounded in efficacy studies, AI benchmarking, and implementation science, creating a framework that connects rigorous research with real-world classroom use. 
What past highlights of yours inform your current approach?
My scientific career began at the Institute for Evidence-Based Education at Southern Methodist University in Dallas, where I worked on large-scale randomized controlled trials of reading and math interventions. That experience shaped my view of research as something that must be systematic, data-driven, and designed to improve outcomes for all children. Later, at Tufts University, Boston Children’s Hospital, and MIT, I trained as a cognitive neuroscientist studying how the brain supports reading and learning.
My research has focused on identifying cognitive and neural risk indicators for dyslexia and understanding how the brain develops the capacity for literacy. We found that early risk markers are both measurable and stable, which strengthened our advocacy for early identification and intervention. Although most states now require universal literacy screening, our recent national survey of educators revealed that many schools still treat screening as a compliance exercise rather than a tool for instructional change. Educators value screening but often lack the training and resources to use the data effectively.
These experiences showed me how individual differences shape learning and how early, evidence-based support can transform outcomes. They also revealed how urgently education needs accountability and transparency. Those lessons now guide EVAL’s mission to connect researchers, developers, educators, and funders to evaluate and improve AI-based learning tools in ways that are rigorous, transparent, and inclusive.
What are your plans and goals with what you are doing through BU?
My research program focuses on understanding how children develop literacy, what happens in atypical development, and how to translate those insights into practical tools for early identification and effective instruction. We are developing AI-based approaches that increase the precision of detecting risk using child speech and creating open-source platforms for scalable, individualized learning. We are also building AI methods that help educators and policymakers make data-informed decisions, particularly to better support students with learning differences. Together, these efforts ensure that innovation in education is guided by evidence and aligned with how children learn.
At Boston University, we recently launched the EVAL EdTech Research Challenge, which offers a free independent evaluation to one company with a promising AI-based learning tool. The response has been inspiring, showing that developers, schools, and funders are eager for more transparent and research-grounded approaches to innovation. Our next goal is to expand the Challenge by partnering with organizations that share our mission and can help us secure the funding needed to provide additional evaluations. In doing so, we hope to shift the incentive structure in educational technology toward evidence, accountability, and real impact on student learning.
‘…we hope to shift the incentive structure in educational technology toward evidence, accountability, and real impact on student learning.’
What are your thoughts on technology’s role in learning, and on AI in education?
I believe in the power of technology and AI to support educators and enhance teaching. Technology can help translate data into actionable insights, generate instructional materials, and identify learner profiles to personalize education. AI-based platforms hold great promise for delivering adaptive, data-informed, and scalable instruction.
The challenge is ensuring that these tools are grounded in rigorous science. We have decades of research on what works, for whom, and under what conditions, yet this evidence is often overlooked in product design and implementation. Just as we would not prescribe medication without testing its efficacy, children should not spend hours with educational tools that have not been proven to improve learning or well-being.
Where do you hope to be with your work in six months to a year?
Over the next year, we plan to complete EVAL’s first independent evaluation and share results that show how rigorous research can guide responsible innovation in education. The strong response to the EVAL EdTech Research Challenge shows there is real appetite for transparent, evidence-based evaluation among developers, educators, and funders.
Our priority is to grow this effort by partnering with organizations that share our mission and can help secure the funding needed to offer more free research evaluations. This support is critical to scaling the Challenge and shifting the incentive structure in educational technology toward evidence, accountability, and meaningful impact. We are building a model where rigorous evaluation drives innovation and raises the standard of both practice and proof in K–12 learning.
At the same time, we continue advancing our research on AI-based tools for early risk identification and individualized instruction. By connecting this work with EVAL’s evaluation framework, we aim to strengthen the link between research, innovation, and classroom practice, ensuring that every educational technology contributes to better outcomes for all learners.
‘…we aim to strengthen the link between research, innovation, and classroom practice, ensuring that every educational technology contributes to better outcomes for all learners.’
Any thoughts on the future of edtech and education?
I am cautiously optimistic about the future of educational technology. There are two possible paths ahead. One is a transformative path, where innovation in AI and edtech is guided by evidence and individualization, leading to real improvements in how children learn and how teachers teach. The other is less hopeful: that despite major investment and widespread adoption, we remain a decade from now with only 30 percent of students proficient in literacy and math, and with even wider gaps in access and opportunity.
We may need better tools, but innovation must go hand in hand with science, even if progress sometimes slows to ensure quality and effectiveness. The future will depend on whether we can align innovation with research and accountability, and on how meaningfully we involve educators, students, and families in the co-design of these technologies. My hope is that EVAL continues to grow as a trusted national model for building the evidence base we need to ensure AI in education supports every learner, not just the fortunate few.
—
Victor Rivero is the Editor-in-Chief of EdTech Digest. Write to: victor@edtechdigest.com
0 Comments