How AI is fueling an existential crisis in education

November 6, 2025·40 min·3 clips

Dr. Adam Dubé explains why the idea of digital natives is a myth that misleads teachers about students' tech skills.

Dr. Adam Dubé is a cognitive scientist and educational psychologist at McGill University whose research focuses on how digital tools affect learning and academic performance. He appears on Decoder with Nilay Patel to discuss what the rise of AI chatbots means for education at a foundational level. Dubé argues that AI has surfaced an existential crisis that was already latent in schooling—namely, that educators have never fully agreed on whether school is about acquiring knowledge, developing cognitive skills, or credentialing students for the workforce. The arrival of tools that can complete assignments instantly has made that ambiguity impossible to ignore. Dubé introduces the concept of the "digital native" as a myth: the idea that younger generations are inherently tech-savvy because they grew up surrounded by devices. His research shows this familiarity with consumer technology does not transfer to productive academic use of the same tools. A significant section covers calculator research: studies tracking elementary students over time find that consistent calculator use correlates with lower math performance, not because calculators are harmful in themselves, but because removing the struggle of calculation also removes the cognitive load that builds number sense. Dubé applies the same framework to AI—when a student delegates the writing process to a chatbot, they bypass the effortful thinking that produces learning. Survey data Dubé cites shows roughly 80 percent of students report using AI to have concepts explained to them, which he views as relatively productive use. Only about 10 percent admit to having AI generate complete assignments, though he acknowledges self-reporting likely undercounts this. The episode explores how teachers are experiencing AI asymmetrically: some educators find that AI dramatically reduces time spent on lesson planning and grading feedback drafts, freeing them for more direct student interaction. Others report that new AI-detection requirements, honor code documentation, and policy enforcement have added administrative burden without any time savings. Dubé discusses the difficulty of designing assessments that are both AI-resistant and still meaningful as learning measures. He notes that oral exams and process-based portfolios are gaining interest but are resource-intensive to implement at scale. The conversation touches on equity concerns: students with better prompt literacy and device access gain a larger productivity advantage, potentially widening existing achievement gaps. Dubé is careful to avoid both techno-utopian and techno-panic framings, describing AI as a tool whose impact depends entirely on how it is embedded in pedagogical practice. He calls for teacher professional development centered not on blocking AI but on helping educators understand what cognitive processes specific assignments are meant to build, so they can make informed decisions about when AI assistance is appropriate. The episode closes with Dubé acknowledging that the field does not yet have longitudinal data on how sustained AI use from early grades affects long-term learning outcomes, making current policy decisions necessarily provisional.

As heard by us

A clear, uneasy look at how generative AI is forcing schools to rethink what education is for.

Decoder with Nilay Patel starts with the obvious classroom fight over students using ChatGPT to cheat, then moves into the stranger problem underneath it: schools are being asked to define what learning is while generative AI is already being sold into lesson planning, grading,…

Read the full review in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Decoder with Nilay Patel