Thinking Ahead: AI, Math, & Minds

Earlier this month author and journalist Michael Pollan was asked what it means to be human in an age of rapidly expanding artificial intelligence. He responded with a warning that “consciousness is under siege,” describing this moment as a re-situating of ourselves as science expands the definition of sentience in the natural world on one side and AI expands on the other. He makes a compelling argument: AI doesn’t know the world, it knows only human representations of the world, and it lacks the embodiment that lies at the root of human consciousness. 

Sitting in the audience at the San Francisco City Arts & Lectures conversation, I stepped away with a question I haven’t been able to shake: As AI tools take on more of the work we used to do ourselves, including in classrooms, what are we at risk of offloading along with that work? Not just calculation but the cognitive struggles that build deep understanding; not just procedures but the habits of mind that lead us to think critically, act with agency, and participate fully in the world. 

Since these tools are already here, modernizing math education to meet this moment is both necessary and urgent. AI is in our pockets, our search bars, our keyboards. Predictive text attempts to complete our sentences (including this one) before we finish them. Google generates AI summaries at the top of our search results, short-circuiting the act of reading and evaluating sources. Many of these tools are so seamlessly woven into daily life that students don’t register them as AI at all, but as convenience. But innovation without reckoning what we may lose—agency over our own minds, our own consciousness—is not modernization. 

And it raises an equity issue: Students who have been historically left behind by our education system—low-income students, students of color, English language learners—may find their education supplemented by AI tools before they have developed the AI literacy to use those tools with intention. That dependence wouldn’t just limit how they use AI, it would also limit the development of the critical thinking skills they need to be agents of their own futures. We don’t yet know all the implications of these tools, but we’re making decisions anyway, and that’s precisely the problem.

As Conrad Wolfram has argued, our math curriculum is far out of step with the world computers have made. He argues that computers don’t just mechanize calculation, they have transformed what mathematics is for. At the same time, our curriculum, especially by the third and fourth years of high school, is something close to 80 percent misaligned with what students actually need for college and career readiness. Wolfram is clear that this isn’t a case for less rigor but a case for applying rigor to the right things and for applying higher-level mathematical thinking to real-world problems. 

As always, history offers a warning. After calculators entered classrooms in the mid-1970s, research generally showed that they didn’t harm learning and, in many cases, helped improve outcomes. But the promise that freeing students from computation would naturally elevate their thinking was assumed rather than designed. A 2024 study of more than 31,000 college students found that 40 percent scored below proficient on critical thinking assessments and that a substantial number graduated without measurable gains in their critical thinking or complex reasoning. The lesson here is that we can’t afford to be passive when it comes to AI, a far more powerful tool that is changing our world faster than we could have imagined—nor can we assume that it will solve long-standing challenges within education.

In fact, we’re starting to get a clearer picture of what happens to our cognitive skills when we start to adopt these new tools and ignore tradeoffs. In 2025, the MIT Media Lab tracked 54 participants over four months using an EEG to measure brain activity while they wrote essays using AI, a search engine, or on their own. The results were striking: Students who used AI consistently showed weaker cognitive engagement, lower memory retention, and a diminished sense of ownership over their own work. If we offload cognitive work before students build the foundation that makes offloading safe, what kind of minds are we developing?

This is where Pollan’s framing, Wolfram’s writing, and the MIT research converge. The issue isn’t necessarily the tools themselves, it’s whether or not human beings remain in the driver’s seat. Instead of asking the question—should students use AI?—we need to be asking what students need to know, deeply and fundamentally, to remain agents in a world that is becoming saturated with these tools and how can we ensure they do.

To be sure, there is a possibility that some of these tools could democratize education by taking on rote tasks, providing scaffolds to students’ own conceptual reasoning and ambitious thinking, and ultimately lowering barriers to STEM. But if AI is used as a replacement for thinking or building relationships with teachers, particularly for underserved students, these technologies risk compounding the inequities they promise to resolve. 

This is a moment for math education, particularly given its history as a gatekeeper, to become a guiding force in ensuring students become critical and creative thinkers with the capacity to act as agents of their own futures, to reach their goals, and to participate in the world with the full power of their minds.