Q&A With Laurence Holt on How AI “Unlocks” New Possibilities in Education
At Renaissance Philanthropy, we believe the rise of artificial intelligence marks one of the most important opportunities in a generation to reimagine how teachers teach and young people learn. Our AI and Education program is focused on identifying and supporting breakthrough ideas that can help every student succeed.
However, few education experts are fluent in both AI and learning, and the data and benchmarks needed to develop sound learning tools are limited. Public R&D investment is small, and most private investors chase scale or profit rather than learning outcomes. We believe that philanthropy can play a unique role by backing ideas that combine technical rigor with educational impact – ideas that the market might otherwise ignore.
Laurence Holt’s AI Unlocks project fits perfectly into that vision. By curating dozens of concrete, high-impact use cases for AI in education, AI Unlocks offers a roadmap for AI researchers and technologists, ed tech developers, funders, and policymakers, who want to act now. It’s exactly the kind of catalytic thinking our field needs because it bridges the sense of opportunity of AI and hard-to-untangle implementation questions to ensure AI’s potential truly benefits teachers and students.
Laurence has spent two decades leading innovation in K-12 education in both the nonprofit and for-profit sectors. He is Senior Advisor at the XQ Institute and formerly served as Chief Product Officer at Amplify and Wireless Generation, overseeing engineering, design, and product development. In this conversation, we discuss AI Unlocks, and why this moment feels different from past waves of ed tech, and how AI could reshape both teaching and learning.
– Kumar Garg
Kumar Garg: What sparked the idea for AI Unlocks?
Laurence Holt: A few years ago, I realized I was often hearing or thinking of ideas that made me say, “Whoa! Someone should build that.” I’ve spent most of my career trying to amplify what works in classrooms, and when AI arrived, it reignited my optimism that technology might finally change how learning happens. But I also know that the market doesn’t always solve education problems.
At first, I just kept a Notion page where I’d jot down these ideas – small, concrete things, made with AI, that could move the field forward. Over time that list grew to around 50 items. Some have since turned into projects, but most remain unbuilt. AI Unlocks is simply a public version of that list – a place where people can discover opportunities, connect, and maybe breathe life into one of these “someone should do that” ideas.
Kumar: So it started as your own scratchpad of what’s possible, and now it’s an open invitation to others to pick up those ideas. What do you hope the site accomplishes?
Laurence: Education doesn’t really have a place like Xerox PARC (Palo Alto Research Center) – a true R&D lab where bold ideas get tested without the pressure of commercial traction. Most startups have to focus on raising money and showing user growth, which makes it very difficult for them to explore ideas that are foundational but not yet products.
AI Unlocks is meant to fill a bit of that gap. Each idea is something that, if built, could enable a whole new class of innovations – like “unlocking” a new door. For example, what if we had a dataset of great teachers supporting a group of students thinking through a topic? You could train AI to support many more groups where the teacher isn’t available. Or what if we had a library of interactive widgets for math and science that AI could control? All the boring chat-based AI tutors would come alive. The hope is that researchers, entrepreneurs, or funders will see an idea and decide to make it real.
Kumar: That makes sense. But people have said for decades that technology would revolutionize education, and the results have been slow. Why might AI finally be different?
Laurence: I often ask people what they think is the probability that AI will bring genuinely transformative change to education, versus incremental improvements. My own estimate is 30 percent – up from maybe 15 percent two years ago. It’s still an uphill climb, but every “Unlock” raises those odds. Too many good ideas and strong research never reach classrooms.
For example, we know that deliberate practice improves learning – the same way you get better at basketball by shooting more hoops. And yet, if you go into most classrooms, you will not find much practice going on. There's a lot of teacher-led instruction, there’s some discussion. Maybe there's practice homework, but that may be shrinking too because the attention attractors outside school are so strong.
The question is, when you add AI into the mix, could those evidence-based approaches, like practice or differentiation become easier to implement and therefore more widespread? AI can help teachers design project-based learning, generate lesson materials, and give students targeted feedback without adding to teachers’ workloads. In short, it can make good practice easier to do. That's the dream at least, and we are seeing a little bit of that starting to happen.
Kumar: You’re describing a pretty fundamental shift. Do you think AI will change teaching more, or learning more? Which will AI have the biggest impact on?
Laurence: It'll change how students learn for sure. We know a lot about learning, and there's not really much magic to it. You have to put in the work, you need feedback, and if you're stuck, you need help, you need support. Historically, a human provided that. Now, AI is pretty good at debugging your math homework, say, spotting where a student’s logic goes awry and nudging them toward the right answer, the way a great teacher would.
When I go into classrooms where they're using AI, I ask students, “Do you want AI to go away? Would you rather go back to how it was?” They say no because they get more time with their teachers. Which if you think about it, it's counter-intuitive. I gave you AI, and you like it because you get more time with your human teacher.
Automation can actually increase human connection in the classroom when it’s used well. The role of the teacher, though, may start to change. Perhaps they become more like a coach: setting goals, motivating students, and helping them overcome roadblocks rather than delivering instruction.
Kumar: That balance between technology and humanity is interesting. Some parents, though, worry about even more screen time. How do you see AI fitting into healthy classrooms?
Laurence: Right. No-one wants to walk into a classroom and see rows of silent students staring at screens. That's not school. What we lack are AI systems that support social learning. For example, helping two students collaborate productively on a problem or guiding small-group discussions.
Right now, when a teacher says, “Turn to your partner and discuss,” the results can be uneven. AI could listen, gently nudge the conversation, and keep both students on task. We don’t yet have that capability, but some of the “Unlocks” on the site focus exactly on this, collecting data and training models to enable more social, less isolating uses of AI.
Kumar: Suppose you could design a new school from the ground up. How would AI fit in?
Laurence: Feedback would be at the center. Students learn fastest when they receive immediate, high-quality feedback, but teachers can only give so much of it. AI can fill that gap the way a coach helps you refine your form shot by shot.
AI can also provide explanations, create animations or interactive examples, and even adapt lessons into different languages. I visited a classroom in the Bronx where newcomers spoke only Fula, and GPT could already communicate with them in their native language, so they were getting feedback on their work where previously they got none. That’s extraordinary.
But AI can’t check students back in, or build community, or celebrate their successes. That’s what great teachers do. So I imagine schools where AI handles the mechanics of instruction and teachers focus on relationships, projects, and purpose. The number of adults in the building wouldn’t shrink, their roles would just become more meaningful.
Kumar: Who do you hope will use AI Unlocks?
Laurence: Anyone who wants to move good ideas from talk to action – entrepreneurs, funders, educators, or researchers. Some foundations are too large to back smaller $100,000-scale experiments, even when those experiments could unlock much bigger opportunities. I hope AI Unlocks helps direct attention to those smaller but crucial gaps.
If you have an idea that belongs there, you can submit it directly through the site. I review each one before posting, and contributors can also flag existing projects that fit a listed unlock. The dream is that, over time, the site becomes a kind of “wisdom of the crowd” for AI in education, a sort of living map of what’s possible and what still needs to be built.
Visit AI Unlocks to explore the ideas or propose your own.