The Take Back Your Brain Campaign continued its campus event series with a roundtable on AI, learning, and cognition, bringing together faculty speakers from different disciplines to consider how generative AI reshapes not only what students produce, but also how they think and learn. Like the previous events in the series, the roundtable aimed to give students, faculty, and staff a few frameworks for thinking critically about AI so they could make more intentional decisions about how, and whether, to incorporate these tools into their lives.
Biology professor Jennifer Ross-Wolff opened the conversation by focusing on what learning requires from the brain. Drawing on the idea of the zone of proximal development, or the “just right learning zone,” she explained that students learn best when a task is neither too easy nor too hard. If it’s too easy, people may disengage. If it’s too difficult, however, stress can make it harder to encode information. She discussed that learning requires both support and challenge, along with quite a bit of practice, repetition, and connection to prior knowledge.
Cognitive science professor Jay McKinney followed by asking how AI interfaces with us as a technology. They introduced the idea of the extended mind, associated with philosophers like Andy Clark, which suggests that tools can become part of our cognitive systems by extending memory or reasoning. McKinney then discussed theories of cognition that understand the mind as shaped by the broader environment, not just by “mind-y” technologies like computers or notebooks. From this perspective, AI certainly can’t be understood as just another tool like a calculator. It has the potential to interfere with the relationship between students and the world, especially when it disrupts the social interactions, uncertainty, and skill-building through which learning happens.
Finally, George Cusack, Director of Academic AI Initiatives, discussed the design of generative AI tools themselves. He explained that these tools are designed to know things and do things so users don’t have to. On the other hand, learning fundamentally requires knowing and doing things for yourself. He made an important distinction between tools that are useful and tools that are helpful. A ladder may be useful for getting up a climbing wall, but not helpful if the goal is to learn how to climb. He also discussed features like instant suggestions, summaries, and refinement buttons as examples of the AI industry’s effort to make these tools feel as seamless as possible, removing the uncertainty and friction that humans often dislike but that are important for learning.
After the presentations, attendees broke into small discussion groups, giving participants time to respond to the speakers’ ideas and reflect on their own relationships to AI and learning.