Dr. Panayiota (Pani) Kendeou is a Distinguished McKnight University Professor and Guy Bond Chair in Reading in the Department of Educational Psychology at the University of Minnesota. She directs the Reading + Learning Lab and co-directs the CEHD Learning Informatics Lab, focusing her research on reading comprehension, digital and media literacy, misinformation, and learning with AI.
In this Q&A, Dr. Kendeou discusses what critical thinking requires in the age of artificial intelligence, why domain-specific knowledge matters, and how education can support the development of expertise.
Reboot Foundation: You have an extensive background in reading comprehension, media literacy, and misinformation. How does your work approach the intersection of critical thinking and AI?
Dr. Kendeou: My research is interdisciplinary, bringing critical thinking into areas such as reading comprehension, digital and media literacy, misinformation, and educational technologies, including AI. Although the literature on critical thinking includes many models and theories, our work focuses on four core components: it needs to be purposeful and reflective, grounded in reasons and criteria, and directed toward good judgments and decisions. These components provide a foundation for examining how people engage with information across different contexts.
Reboot Foundation: What does critical thinking look like when we interact with generative AI?
Dr. Kendeou: In our work on reading and media literacy, critical thinking involves evaluating claims, arguments, evidence, and sources to make well-justified decisions. But it is also situational: it depends on the task and on what the person knows. That matters for both education and workforce development. If we want to use AI to develop critical thinking, we need to consider the knowledge and skills learners bring to the interaction. Without a sufficient foundation, they may struggle to evaluate what AI produces, putting the development of critical thinking at risk.
Reboot Foundation: Does AI fundamentally change how we evaluate information, or are we using the same skills, just applied to a different medium? For example, 30 years ago, if I was thinking critically about a magazine article, how would that be different from critically thinking about an AI response today?
Dr. Kendeou: The core skills should be the same: evaluating sources, arguments, evidence, and claims. What changes is the context in which we apply them. Unlike a static article, generative AI allows for a back-and-forth exchange, with responses that can adapt to the user. That interaction may also shape perceptions of the system’s expertise and trustworthiness. We still need to understand how these features affect critical thinking. The key question is how people apply familiar evaluative skills in a context that is interactive, adaptive, and personalized.
Reboot Foundation: When looking at the current resources available for teaching critical thinking with AI, where are the biggest gaps?
Dr. Kendeou: There is plenty of guidance on using AI, but I would like to see a more systematic effort to integrate critical thinking into specific subject areas. Its core components apply across domains, but using them depends on knowledge within those domains. We have seen work in reading, mathematics, and other STEM fields; the challenge is to make this integration intentional and sustained. My concern is that critical thinking could be treated as an add-on, as media literacy sometimes has been, rather than as part of everyday instruction. It needs to be embedded in reading, writing, and STEM, where students can develop and apply it through the content they are learning.
Reboot Foundation: Since critical thinking is highly situational, what does it look like when embedded into a specific domain, like reading?
Dr. Kendeou: We can build a progression from a single text to multiple texts and sources. With one text, students can evaluate claims and reasons, identify misinformation, and examine refutations and explanations. As they move to multiple sources, they encounter converging or conflicting evidence and plausible alternatives. They then need to comprehend and synthesize that information to make justified judgments. Our theoretical models of reading have evolved to account explicitly for these processes and decisions. That allows us to test the models and use them to inform instructional practices that help students develop these skills.
Reboot Foundation: What is at stake if young professionals rely on AI before developing foundational knowledge and skills?
Dr. Kendeou: My concern is that we could undermine the development of expertise. Foundational knowledge is essential to becoming an expert. If reliance on AI prevents people from developing that knowledge, the consequences extend beyond their immediate performance on a task. There is a tendency in public discourse to equate access to information with expertise. But being able to find information is very different from having the knowledge needed to understand, evaluate, and use it. That distinction needs to remain central as we bring AI into education and professional learning.
Dr. Kendeou recently led the development of a new examination of how generative AI is both aiding and complicating the learning experience for students of all ages.
