In today's rapidly evolving educational landscape, Artificial Intelligence is emerging as a transformative force, offering both opportunities and challenges. As AI technologies continue to advance, it's crucial to examine their impact on student expectations, learning experiences, and institutional strategies. One pressing question is: what do students truly want from AI in education? Are they reflecting on the value of their assessments and assignments when AI tools can potentially complete them? This begs the deeper question of what we mean by student success in higher education and the purpose of knowledge in an AI-driven economy.  Professor Rose Luckin is joined by three wonderful guests in the studio to discuss what tools we need to support students and how we explore the potential and the limitations of AI for education. Guests: Michael Larsen, CEO & Managing Director, Studiosity Sally Wheeler, Professor, Vice-Chancellor, Birkbeck, University of London Ant Bagshaw, Executive Director, Australian Technology Network of Universities Talking points and questions include: Student expectations and perspectives on using AI for assessments/assignments and the role of knowledge in an AI economy The potential of AI to enhance learning through features like instant feedback, error correction, personalized support, learning analytics How AI could facilitate peer support systems and student community, and the research on the value of this The lack of robust digital/AI strategies at many institutions as a barrier to effective AI adoption The evidence-base for AI in education - challenges with research being highly specific/contextual, debating the value of in-house research vs general studies Whether evidence on efficacy truly drives institutions' buying decisions for AI tools or if other factors/institutional challenges are stronger influences How challenges facing the education sector can inhibit capacity for innovative deployments like AI The growing need for proven, supportive AI tools for students despite institutional constraints