A $750,000 National Science Foundation grant will fund Cornell researchers studying whether generative AI can help students build critical thinking skills instead of undermining them. The three-year project, led by biomedical engineering assistant professor Alexandra Werth, aims to develop an AI tool that prompts students to reflect on daily lectures and labs - the kind of personalized interaction a human instructor would normally provide.
The tension sits at the center of the project. Students can use AI tools to bypass the reasoning processes that develop critical thinking. At the same time, because generative AI produces inaccurate or fabricated information, students need sharper critical thinking to question, verify, and decide what to trust.
The tool and the question
The Automated Critical Thinking Reflection (ACTR) tool uses generative AI to deliver personalized reflection prompts. Werth, the principal investigator, frames the problem directly. "In the age of generative AI in education, students' development of critical thinking skills is a major concern," she said. "Students can use these tools to circumvent or outsource the reasoning processes through which critical thinking skills are developed, allowing generative AI to make decisions for them."
Werth added that the same technology that creates the problem might also address it. "Since generative AI can produce inaccurate or fabricated information, students need critical thinking to determine what to question, what to verify, what to believe and how to act." The project will test that proposition across physics, biology, and engineering courses.
Who's involved
Co-principal investigators include Natasha Holmes, the Ann S. Bowers Associate Professor of physics in the College of Arts and Sciences, and Michelle Smith, Distinguished Professor of Arts & Sciences and Senior Associate Dean for Undergraduate Education. All three are discipline-based education research faculty members - scientists who study how students learn within their own fields.
The interdisciplinary scope spans courses in physics, biology, and engineering. Researchers will evaluate the role of AI for education research while simultaneously using the AI systems to build student reflection skills. For educators tracking how AI enters the classroom, this dual-purpose approach mirrors broader questions in AI for Education: whether the tools that threaten certain skills can also strengthen them.
What the grant supports
The NSF funding enables the team to build and assess the ACTR tool across multiple semesters and disciplines. Rather than studying AI in the abstract, the project embeds it directly into daily lecture and lab reflection - the routine moments where critical thinking either develops or gets skipped. Faculty interested in similar approaches can find structured guidance through AI for Teachers learning paths designed for classroom integration.
Why this matters for education professionals
This project addresses a practical problem every instructor now faces: students are using generative AI, and blanket bans won't work. The research tests a specific alternative - using AI to prompt the very reflection that AI otherwise short-circuits. For teachers and administrators making decisions about classroom AI policies, the findings will offer evidence on whether structured reflection tools can turn a threat to critical thinking into a scaffold for it. The results won't arrive immediately, but the questions the Cornell team is asking are the ones most departments are already wrestling with.
Your membership also unlocks:
