Generative AI can do a remarkable amount of the work we have traditionally asked learners to do. It can answer questions, summarize texts, analyze sources, construct arguments, and make decisions. As these technologies become increasingly embedded in education, an important question to consider is where the thinking happens. What happens when AI makes difficult intellectual work easier? Which parts of learning are too important to give away?

Drawing on work in history and social sciences education, this keynote uses technocuriosity to think beyond the familiar choice between embracing AI and resisting it. Technocuriosity invites us to play with and explore new technologies while also asking questions about how they work, what they make possible, and what may be lost along the way. History education offers a particularly useful place for thinking through these questions because learning history requires more than arriving at an answer. Learners encounter incomplete and conflicting evidence, consider perspective and context, wrestle with uncertainty, and make judgments about what the available evidence allows them to say.

Using examples from work with historical inquiry, primary sources, questioning, and AI-mediated learning scaffolds, we will explore what it might mean to design encounters with generative AI that keep learners engaged in the difficult work of wondering, questioning, interpreting, revising, and making judgments. As AI increasingly shapes how we encounter information and make sense of the world, these questions can guide what it means to learn, participate, and exercise agency in an AI-enabled society. The challenge extends beyond deciding whether AI belongs in education. It is to figure out what the “good stuff” of learning is, and how we make sure learners still get to do it.