Terence Tao on managing student AI use: struggle builds skills

Terence Tao, speaking as part of the 2026 EMS Lecture Series on Mathematics Education, presented a nuanced case for how university students should manage their relationship with AI tools. Rather than viewing AI as either a panacea or a threat to be banned, Tao frames the issue in terms of a "cognitive diet"—drawing parallels to nutrition and physical exercise to argue that the challenge of the current era is managing abundance rather than scarcity.

Tao identifies four concrete risks of unrestricted AI use. First, deskilling: over-reliance on AI leads to the atrophy of foundational problem-solving and critical thinking abilities. Second, learned helplessness: students become psychologically dependent, feeling unable to start or solve problems without AI assistance. Third, loss of cognitive diversity: AI models tend to converge on mainstream, popular, or repetitive answers rather than generating the unique, diverse, or organic insights that emerge from human struggle. Fourth, sycophancy: AI's tendency to prioritize affirmation over accuracy undermines the corrective feedback that drives learning.

To counter these risks, Tao recommends three shifts in educational practice. First, de-emphasize the correct answer and instead focus on the process of discovery and verification—the journey rather than the destination, which AI can trivialize. Second, encourage "productive failure": normalize the experience of struggling or failing first, since this is where genuine learning occurs. Teachers should design tasks where AI assists or is coached rather than replacing student effort. Third, use AI sparingly to present concepts in unusual or creative formats to maintain engagement and avoid monotonous textbook repetition.

Central to Tao's proposal is a "Blue Team vs. Red Team" framework. Students should use AI for creative production (Blue Teaming) only if they develop the skills to rigorously critique and verify its output (Red Teaming). This is not permission to use AI freely; it is a competency requirement. Students must be capable of explaining and justifying their use of AI tools in class. In essence, Tao argues for a hybrid approach where AI is treated like a vehicle meant for specific tasks—freeing up time and effort—while students continue to "walk" using traditional mental effort to maintain core intellectual fitness. The underlying message is that open, transparent dialogue between educators and students about the healthy use of AI tools is essential as institutions navigate this transition.

Tao's lecture reflects a broader institutional tension: as AI tools become more capable and accessible, educators face pressure to either restrict them or integrate them responsibly. Tao rejects the binary of "ban vs. enable." Instead, he frames the challenge as a diet problem—not whether to consume AI, but how much, when, and in service of what educational goal. His parallel to nutrition and physical exercise is instructive: just as muscles atrophy without use, cognitive skills degrade when students outsource thinking to AI before mastering it themselves.

The shift from cognitive scarcity to cognitive abundance is real and documented in the body: tasks that once forced struggle now can be bypassed entirely. Tao's concern is not merely that answers are easier to obtain, but that the struggle itself—the messy, frustrating middle where learning crystallizes—vanishes. This is why he emphasizes "productive failure" and the process of discovery. His Red Team / Blue Team model operationalizes this insight: students earn the right to use AI as a tool by first proving they can think critically about its output. This positions AI not as a shortcut but as an advanced instrument available only to those who have already built foundational skill.

No comments yet. Be the first to share your thoughts!