When social media first entered our world, educators—myself included—focused narrowly on strategies for incorporating these tools into classroom contexts. We were so busy figuring out the “how” of implementation that we paid far less attention to the “what if” of broader societal impact. As I noted in a previous post, there are entries on this very website that stand as testimony to that naivete. I did not spare a thought to the possibility that these technologies might exacerbate polarization, erode trust in institutions, or harm the mental health of our most vulnerable.

These consequences are now evident, forming significant challenges for our educational institutions. As we face the rise of Generative AI, the question is whether we will let history repeat itself or simply rhyme. We do not have the luxury of waiting to find out.

That urgency is the driving force behind a new report released today by the Center for Universal Education at The Brookings Institution: A New Direction for Students in an AI World: Prosper, Prepare, Protect.

I had the honor of contributing to this work as a member of the Global AI Task Force. Unlike traditional studies that conduct postmortems on past failures, our team undertook a global “premortem.” To summarize:

After interviews, focus groups, and consultations with over 500 students, teachers, parents, education leaders, and technologists across 50 countries, a close review of over 400 studies, and a Delphi panel, we find that at this point in its trajectory, the risks of utilizing generative AI in children’s education overshadow its benefits.

What I appreciate most about this work is that it goes beyond generic warnings to identify why these risks are escalating. It focuses heavily on the “instructional core”—the essential human web between student, teacher, and content—arguing that AI currently diminishes learning by weakening these interactions. It warns against “sycophantic” AI (models designed to flatter students rather than challenge them) and highlights the dangers of cognitive offloading, where the tool supplants the thinking process itself.

However, the report argues that this trajectory is not fixed. It calls for “carefully titrated” AI use, using these tools with the precision of a prescription rather than a firehose, and proposes a path forward built on three pillars: Prosper, Prepare, and Protect. Rather than just adopting tools, we need to design educational environments where AI enriches rather than diminishes human capacity, building true capabilities for an AI-powered world while rigorously safeguarding student well-being.

You can read the report or versions of it by following the links below:

• Full report A New Direction for Students in an AI World)
• Summary
• One pager

I want to extend my gratitude to the lead authors: Rebecca Winthrop, Mary Burns, Natasha Luther, Emma Venetis, and Rida Karim for orchestrating and conducting the work that went into this. And to all the members of the task force for their rigor and intellectual honesty. It has truly been a pleasure to work together with this amazing team on this important report.