Dear readers,
Can you believe we are already a full month into 2026? Almost everyone I know is feeling more than a little overwhelmed by the realities of catching up on work after the holidays. And then there’s the news cycle and what’s happening in Minneapolis: the human toll, the intense levels of stress and disruption among families, and the fact that growing numbers of kids are not in school are on all of our minds, I’m sure. This Beth Hawkins piece in The 74 about the impact on the children of Minneapolis is sticking with me.
In times of overload, I always find solace in ideas. In that spirit, I’ll zero in on just one big idea this edition.
A call for coherence and ambidextrous thinking in approaching AI in education
Our Think Forward: Learning With AI Forum paper is out. It captures the collective wisdom of our first Think Forward Fellowship Cohort—a group of thoughtful and experienced leaders from education, advocacy, ed tech, policy, research, and philanthropy. Many were AI skeptics. They came together under the assumption that the education sector must move quickly to act on the opportunities and navigate the challenges AI presents. As one fellow put it:
This past November, forty of us gathered in Albuquerque, New Mexico, with the Sandia Mountains as our backdrop and an impossible question in front of us: How should education leaders navigate AI when the technology is moving faster than policy, faster than pedagogy, and faster than our capacity to make sense of it all?
The group sounded an alarm over the “efficiency paradox” driving AI adoption right now. The tech space is flooded with point solutions aimed at maximizing productivity gains, like saving teachers’ time. While our fellows recognize that teachers need relief from the most overwhelming and mundane parts of teaching, true efficiency is not just about time saving; it’s about maximizing productivity while reducing costs. Right now, we’re nowhere near that when it comes to AI in education.
Fellows also warned that AI tool adoption and use cases are too often fragmented and misaligned with learning science and coherent whole-school design. This prevents educators from truly delivering on the promise of Gen AI to personalize and accelerate student learning.
The group also felt strongly that AI tools must be in service of a broader vision for future-ready education, one that recognizes and addresses the true pain points in education now and into the future. Kids don’t like school, often finding it boring and irrelevant; current staffing models are broken; mental health issues are profound. At the same time, the world is changing dramatically, and we are not talking enough about how to truly prepare kids for what is coming.
The meeting surfaced a critical tension in the group (and one that I suspect is emerging more broadly among funders and advocates): Should AI solutions focus on making our current education systems work as well as possible? Or should attention and resources go to building a new and completely different educational model instead?
After much debate, fellows called for an “ambidextrous” approach to funding, policy, and practice. They acknowledged that we must help as many students as possible in the current structures. But they also issued an urgent call to action to move toward future-ready schooling, including investments in new whole-school designs, “policy stacks” that incentivize and support moves toward things like mastery-based learning and new approaches to staffing schools. These new staffing approaches would fully integrate AI into team-based teaching, teacher specializations, and new ways to leverage community-based experts. The group was thinking big, but they were also pragmatic and talked about how to use AI to power existing programs like CTE and other career pathways initiatives.
One of my favorite outputs from the group was this list of principles that could guide a diverse array of future-ready school models:
• Human first: AI should be a catalyst for human-centered learning, not a replacement.
• Purpose over efficiency: Focus on education as a step towards meaning-making, curiosity, and compassion, all human emotions in a world of artificial intelligence.
• Students and educators at the center: System design and resources must be focused on students as active learners and decision-makers, with educator roles changed to better meet their needs.
• Equity and ethics by design: Guardrails and policy are at the forefront of protecting students and assuring access, not an afterthought.
• Community-rooted and globally connected: Educators should partner with families and communities to connect learning to students’ local contexts and broader global opportunities.
• Coherence beats novelty: AI solutions should reinforce each other and collectively advance a clear vision for student outcomes and learning.
• Transparency and real-world application: Help families understand how learning connects to careers, college options, and the long-term outcomes of different educational pathways.
• Continuous lifelong learning: Power adult learning beyond K-12 through dynamic, adaptive, and data-driven systems.
• Infrastructure for liberation: AI should operate in the background to reduce friction and expand the capacity of students and educators.
• Rigorous personalization: Grounded in the science of learning and refined for the age of AI.
We’ll also be publishing some of the whole school design visions that Fellows developed. Exciting stuff.
In any event, there is a LOT of provocative thinking in the report. See my colleague Maddy Sims’s op-ed about myths versus facts and a really thoughtful reflection from one of our fellows, Nick Potkalitsky. My hope is that this paper encourages even more reflections and debates in the field. I’ve already heard from one friend who didn’t think the paper went far enough in its recommendations and offered his own ideas. This is exactly the kind of response we want. I’ll be writing more soon about the feedback we’re getting and the ideas the paper is generating…so send me your thoughts.
Take good care and more soon on Moltbook (a new social media network for AI agents?!) and more.
-RL