1. You've spent a decade evaluating education innovations as an investor, and now you lead the Stanford Accelerator for Learning. When it comes to building EdTech products that genuinely reflect what learning science tells us, what does that process look like when it's done well?

‍

The biggest mistake is treating learning science as something you consult at the end of product development. It needs to be there from day one. The best teams start with a learning problem, not a technology. They bring together researchers, educators, designers, students, and engineers from the outset. They prototype, test, measure, iterate, and remain humble enough to change course when the evidence tells them to.

‍

At the Stanford Accelerator for Learning, we often say that breakthrough educational innovation happens at the intersection of rigorous research, thoughtful design, and real-world implementation. When those three stay connected, technology becomes much more than software. It becomes a catalyst for better learning.

‍

2. You talk about relational intelligence as the skill of the AI era. For an EdTech product team sitting down to build something, what does it actually look like in practice?

‍

‍

One of the strongest findings from the learning sciences is that we don't simply learn alone, we learn with and through other people. Relationships shape not only what we know, but who we become. Relational intelligence (RQ) is the capacity to build trust, communicate across differences, collaborate, repair conflict, and create belonging. As AI becomes increasingly capable, these human capacities become even more valuable. So I would ask every product team one simple question: Does your product strengthen relationships, or quietly replace them?

‍

The most exciting AI products don't simply personalize learning for individuals. They help teachers know students better. They spark richer conversations among peers. They strengthen partnerships with families. They give educators more time for mentoring, coaching, and human connection. Technology should amplify relationships, not compete with them.

‍

‍

3. You've referenced the AEMS framework — Active, Engaging, Meaningful, Social — as a way to evaluate whether technology is helping or getting in the way. If an EdTech company or an institution applied that filter honestly to their current tools, what do you think they'd find?

‍

One of the most influential contributions from my dear colleague Kathy Hirsh-Pasek is the AEMS framework: learning should be Active, Engaged, Meaningful, and Social. I think every education company should use it as a design checklist.

‍

If many products applied that filter honestly today, I suspect they'd score well on engagement. AI is becoming remarkably good at capturing attention. Far fewer would score highly on meaningful learning, or especially on the social dimension. We've spent years optimizing learning for individuals, yet many of the capacities that matter most for life develop through interaction with other people. Think collaboration, empathy, leadership, communication, or even curiosity and creativity.

‍

As I think about the future, I sometimes wonder whether we should modify the last letter or add a new one: “R” for Relationships. Not because it replaces Kathy's framework, but because AI raises a new design question: Does this technology strengthen human relationships, or replace them? That may become one of the defining questions for education in the AI era.

‍

4. The evidence for investing in early childhood is overwhelming. You've seen this from the philanthropy side, the investment side, and the research side. How do you read the current state of funding and EdTech development in that space, and what would it take to match the level of investment to the evidence?

‍

The science has been remarkably consistent for decades. Early relationships shape brain development, language, executive function, mental health, and lifelong learning. We know that the earliest years generate some of the highest returns of any educational investment. Yet our investments still don't reflect that evidence. At the same time, we should be thoughtful about where technology belongs in early childhood. Young children learn primarily through responsive interactions with caring adults, not through screens. The goal shouldn't be more technology for children. It should be better technology for the adults who care for them.

‍

I see tremendous opportunity for AI to support parents, caregivers, and educators: helping them understand child development, suggesting age-appropriate activities, reducing administrative burdens, identifying developmental concerns earlier, and giving them more time and confidence to engage in rich, back-and-forth interactions with children.

‍

In other words, the most promising use of AI in early childhood is not to replace human relationships, but to strengthen them. If we design technology that helps adults be more present, more responsive, and more connected to children, we'll be using AI in service of what the science has told us all along: relationships are the foundation of learning.

‍

5. You've judged some of the largest innovation competitions in education. When you're evaluating whether something is worth scaling, what do you look for? What tends to separate the innovations that reach learners at scale from those that don't?

‍

The first thing I look for is whether the team has fallen in love with a technology or with a learning challenge. The latter almost always wins. The innovations that scale solve a meaningful need, demonstrate evidence that they improve learning, fit naturally into educators' workflows, and are designed with the realities of schools in mind. The best founders also listen exceptionally well. They treat teachers, students, and families not as users, but as co-designers.

‍

Ultimately, scale isn't about reaching millions of users. It's about improving millions of lives.

‍

6. There's a version of AI in education that just automates the system we already have, and there's a version that changes what's possible. You've said we risk automating the past. What does the other path look like, and who do you see getting it right?

‍

Every technological revolution forces us to ask what remains uniquely human. AI should prompt the same reflection. One path simply automates yesterday's education system. It makes lectures, worksheets, grading, and standardized instruction more efficient. That is not transformative. The more exciting path asks a different question: If AI increasingly augments cognitive work, what human capacities become even more valuable?

‍

I believe the answer includes curiosity, creativity, adaptability, ethical judgment, resilience, and above all our capacity to build meaningful relationships. I'm encouraged by organizations that see AI as a partner rather than a replacement, and many of the researchers and entrepreneurs working with us at the Stanford Accelerator for Learning. They recognize that AI should empower teachers, not replace them, and expand what is possible for learners.

‍

For much of the last century, education was shaped by what we could measure most easily, primarily cognitive ability. AI gives us an opportunity to broaden that vision. The future of education is not simply about making students more adept at using AI. It's about helping people increase human potential, become wiser collaborators, stronger communities, and more fully human.

If the 20th century was shaped by IQ, and recent decades expanded our focus to EQ, I believe the AI era invites us to cultivate Relational Intelligence (RQ).

Isabelle has spent a career refusing to separate what the science says from what the technology does. 

‍

What we want EdTechs to take away from this conversation: there’s a design question the sector has been slow to confront. Most solutions optimize for the individual learner, but what about collaboration, empathy, trust, belonging?

‍

Isabelle has pointed out how AI makes this harder to ignore. It’s great for education technology to be more personalized, but how can you make it more relational and ultimately more human? 

‍

1. You've spent a decade evaluating education innovations as an investor, and now you lead the Stanford Accelerator for Learning. When it comes to building EdTech products that genuinely reflect what learning science tells us, what does that process look like when it's done well?

‍

The biggest mistake is treating learning science as something you consult at the end of product development. It needs to be there from day one. The best teams start with a learning problem, not a technology. They bring together researchers, educators, designers, students, and engineers from the outset. They prototype, test, measure, iterate, and remain humble enough to change course when the evidence tells them to.

‍

At the Stanford Accelerator for Learning, we often say that breakthrough educational innovation happens at the intersection of rigorous research, thoughtful design, and real-world implementation. When those three stay connected, technology becomes much more than software. It becomes a catalyst for better learning.

‍

2. You talk about relational intelligence as the skill of the AI era. For an EdTech product team sitting down to build something, what does it actually look like in practice?

‍

‍

One of the strongest findings from the learning sciences is that we don't simply learn alone, we learn with and through other people. Relationships shape not only what we know, but who we become. Relational intelligence (RQ) is the capacity to build trust, communicate across differences, collaborate, repair conflict, and create belonging. As AI becomes increasingly capable, these human capacities become even more valuable. So I would ask every product team one simple question: Does your product strengthen relationships, or quietly replace them?

‍

The most exciting AI products don't simply personalize learning for individuals. They help teachers know students better. They spark richer conversations among peers. They strengthen partnerships with families. They give educators more time for mentoring, coaching, and human connection. Technology should amplify relationships, not compete with them.

‍

‍

3. You've referenced the AEMS framework — Active, Engaging, Meaningful, Social — as a way to evaluate whether technology is helping or getting in the way. If an EdTech company or an institution applied that filter honestly to their current tools, what do you think they'd find?

‍

One of the most influential contributions from my dear colleague Kathy Hirsh-Pasek is the AEMS framework: learning should be Active, Engaged, Meaningful, and Social. I think every education company should use it as a design checklist.

‍

If many products applied that filter honestly today, I suspect they'd score well on engagement. AI is becoming remarkably good at capturing attention. Far fewer would score highly on meaningful learning, or especially on the social dimension. We've spent years optimizing learning for individuals, yet many of the capacities that matter most for life develop through interaction with other people. Think collaboration, empathy, leadership, communication, or even curiosity and creativity.

‍

As I think about the future, I sometimes wonder whether we should modify the last letter or add a new one: “R” for Relationships. Not because it replaces Kathy's framework, but because AI raises a new design question: Does this technology strengthen human relationships, or replace them? That may become one of the defining questions for education in the AI era.

‍

4. The evidence for investing in early childhood is overwhelming. You've seen this from the philanthropy side, the investment side, and the research side. How do you read the current state of funding and EdTech development in that space, and what would it take to match the level of investment to the evidence?

‍

The science has been remarkably consistent for decades. Early relationships shape brain development, language, executive function, mental health, and lifelong learning. We know that the earliest years generate some of the highest returns of any educational investment. Yet our investments still don't reflect that evidence. At the same time, we should be thoughtful about where technology belongs in early childhood. Young children learn primarily through responsive interactions with caring adults, not through screens. The goal shouldn't be more technology for children. It should be better technology for the adults who care for them.

‍

I see tremendous opportunity for AI to support parents, caregivers, and educators: helping them understand child development, suggesting age-appropriate activities, reducing administrative burdens, identifying developmental concerns earlier, and giving them more time and confidence to engage in rich, back-and-forth interactions with children.

‍

In other words, the most promising use of AI in early childhood is not to replace human relationships, but to strengthen them. If we design technology that helps adults be more present, more responsive, and more connected to children, we'll be using AI in service of what the science has told us all along: relationships are the foundation of learning.

‍

5. You've judged some of the largest innovation competitions in education. When you're evaluating whether something is worth scaling, what do you look for? What tends to separate the innovations that reach learners at scale from those that don't?

‍

The first thing I look for is whether the team has fallen in love with a technology or with a learning challenge. The latter almost always wins. The innovations that scale solve a meaningful need, demonstrate evidence that they improve learning, fit naturally into educators' workflows, and are designed with the realities of schools in mind. The best founders also listen exceptionally well. They treat teachers, students, and families not as users, but as co-designers.

‍

Ultimately, scale isn't about reaching millions of users. It's about improving millions of lives.

‍

6. There's a version of AI in education that just automates the system we already have, and there's a version that changes what's possible. You've said we risk automating the past. What does the other path look like, and who do you see getting it right?

‍

Every technological revolution forces us to ask what remains uniquely human. AI should prompt the same reflection. One path simply automates yesterday's education system. It makes lectures, worksheets, grading, and standardized instruction more efficient. That is not transformative. The more exciting path asks a different question: If AI increasingly augments cognitive work, what human capacities become even more valuable?

‍

I believe the answer includes curiosity, creativity, adaptability, ethical judgment, resilience, and above all our capacity to build meaningful relationships. I'm encouraged by organizations that see AI as a partner rather than a replacement, and many of the researchers and entrepreneurs working with us at the Stanford Accelerator for Learning. They recognize that AI should empower teachers, not replace them, and expand what is possible for learners.

‍

For much of the last century, education was shaped by what we could measure most easily, primarily cognitive ability. AI gives us an opportunity to broaden that vision. The future of education is not simply about making students more adept at using AI. It's about helping people increase human potential, become wiser collaborators, stronger communities, and more fully human.

If the 20th century was shaped by IQ, and recent decades expanded our focus to EQ, I believe the AI era invites us to cultivate Relational Intelligence (RQ).