“Attention is what taught the machines. But now, in this AI world, we need to have intention and judgment. Intention is what we need.”

I loved that conversation with Neeru Khosla, co-founder of the CK-12 Foundation, a nonprofit focused on free, adaptive learning materials for students and teachers. She articulated one of the most important shifts now unfolding in education. For years, digital learning has focused on access: more content, more devices, more formats, more scale. But Neeru’s point is that the real shift brought by AI is the possibility of understanding how a student is actually thinking – where they are confused, what misconception they hold, what question they are really trying to ask, and what kind of help might move them forward.

Khosla has been working on this problem for nearly two decades. CK-12 started with an idea to make high-quality learning materials free, flexible, and adaptable to the needs of individual students. Long before the current AI boom, the foundation was already building multimodal learning tools, adaptive systems, and concept-based content designed to meet learners where they are. But as Neeru describes it, the missing layer was always insight. Schools could see whether a student got something right or wrong. They could not easily see why.

That is where the last three years changed everything. In Neeru’s view, AI is not most interesting as an answer machine.

“Just knowing that the answer is right or wrong doesn’t tell anybody anything.”

It is most useful when it helps surface the learner’s internal process. A question like “How does the sun burn if there’s no oxygen in space?” is not just a cute moment of curiosity – it reveals a specific misunderstanding that a good tutor can work with. This is the logic behind Flexi, CK-12’s AI tutor, which Neeru says has already been used by tens of millions of students and has processed more than 150 million questions. No one needs automation for its own sake. New tools give us the ability to trace learning more deeply than standardized testing ever could.

But there is a lot of fear in the system. Our conversation moves well beyond product and platform design. We talk about why schools often resist free tools, why they now resist AI tools, why teachers need open-mindedness more than technical fluency, why AI literacy may become as foundational as reading and writing, and why creativity, critical thinking, collaboration, communication, and community remain the real non-negotiables. Khosla also makes a broader argument: education is one of the few places where society cannot afford to treat access as optional. If AI is going to matter in the classroom, it has to help more children learn more deeply – not just help the most privileged students move faster. She argues that if the US wants to maintain its lead, it needs to rethink education and how it is delivered.

There is also something personal and unusually grounded in the way she talks about all this. Neeru came to education from molecular biology, then spent years in classrooms watching children learn one by one before building CK-12. That background gives her a practical lens on both technology and human development. She is enthusiastic about AI, but not dazzled by it. Again and again, she returns to the same principle: tools matter, but what matters more is whether they help us show up – for students, for teachers, and for one another.

Neeru believes in two AIs: Augmented Intelligence and Amplified Intention, and I’m 100 percent agree with her.

As I keep saying, AI literacy is extremely important, and this conversation is a great way to learn how to move from fear to understanding. Please watch it and share it →

We prepared a transcript for reference, but the full experience is in the video. And as always: like and comment. It helps us grow on YouTube and bring you more insights.

Ksenia:
Today I’m very honored to speak with Neeru Khosla. She is an entrepreneur and philanthropist known for her work in education and social impact. She is the co-founder of CK-12 Foundation, which provides free, customizable learning materials to students and teachers worldwide.

I’m very passionate about education, especially in this new era of AI, so thank you so much for joining me.

Neeru:
Thank you for having me.

Ksenia:
This podcast is called Inference – because we’re trying to make sense of what’s happening and draw conclusions from it.

Neeru:
Thank you for inviting me to Inference. I love talking to people about what I care deeply about, because education is one of the most important things we can give back to society. We can create all kinds of financial freedom, but that can only happen if we truly help young students understand what they need to know – and how to use that knowledge.

Education in the AI Era

Ksenia:
The last three years have been pretty disruptive for many industries. How has that changed your perspective on education and on the CK-12 Foundation?

Neeru:
Let me go back a little.

When we started CK-12, we were trying to answer a simple question: how might we make education available and meaningful for all children? At the time, textbooks were very one-sided. You got what you got. But true learning only happens when students can make their own connections to what they are learning.

As technology began to evolve in the early 2000s, I knew it was going to go far. I thought this would be one way to reach many more students and help them. So from the beginning, we made CK-12 free.

Technology gave us capabilities that learning really needs – especially multimodality. You cannot always learn effectively just by reading a textbook or just by watching a video. A concept has to be presented in the modality that works best for the learner.

Take physics, for example. If you are learning about electricity, reading in a textbook about how electrons move doesn’t really make it come alive. Now we have simulations. We have ways for students to actually see and interact with what they’re learning.

That was powerful. But even then, we still didn’t really know what was going on in the minds of students. We didn’t know whether they truly understood what we were trying to teach.

It was only with the arrival of AI in the last three years that we realized there was a real opportunity there. It wasn’t perfect – and no technology ever arrives 100 percent ready. Technology matures through iteration, and that requires all of us to participate.

When OpenAI suddenly reached enormous adoption in a very short time, it was clear something major was happening. A lot of people were afraid: students will cheat, machines will replace us, and so on. I was afraid as well. But I also noticed something important: prompting itself is a learning tool.

Ksenia:
Absolutely. It’s the skill of asking questions.

Neeru:
Exactly. That’s why I kept telling people not to be afraid. If students can prompt a machine in a way that gets them where they want to go, they are thinking. They are thinking about concepts, about application, about whether something sounds right or wrong, about whether they want to go deeper.

They may not describe it that way, but they do know when an answer doesn’t sound right. Then they keep prompting. If we can teach children to question – to keep asking why, the five whys – that itself is a huge learning process.

So over the last three years, we built on top of everything we had already done in learning science, concept maps, and knowledge connections. Two and a half years ago, we introduced Flexi, our student tutor. Today we have over 50 million students using it.

Ksenia:
Wow. And how are the results?

Neeru:
We’ve had over 150 million questions asked.

From the beginning, I told the team that we had to categorize and tag the questions students were asking. Many are procedural: help me solve this, define this, help me understand this. Those are fine.

But the questions that really matter are the ones that reveal thinking. For example: How does the sun burn when there’s no oxygen in space? That’s a wonderful question from a young learner. You immediately see the misconception: the child is assuming the sun “burns” through combustion, when in fact it’s nuclear fusion.

That is what we built: a platform that connects concepts and content. Our FlexBooks were always designed to be flexible and personalized – to each learner’s level, language, standards, and classroom needs. But until AI, we couldn’t really see inside the learning process at this level.

And that changes everything. Standardized testing only tells you whether an answer was right or wrong. It doesn’t tell you why a learner got something wrong. Now we can do deep knowledge tracing and understand where the misunderstanding happened.

Do We Need to Rethink Education From the Ground Up?

Ksenia:
I have five children, and they’re still in public school. I can see how early we still are with AI because public schools just can’t catch up that fast. Do we need to rethink the whole education system from the ground up?

Neeru:
When I started, I actually avoided trying to get adoption directly in public schools. Partly because they didn’t want to adopt something free. We are still free, almost 20 years later, and we intend to remain free.

What surprised me early on was that schools, districts, and states had money – but they didn’t want to lose the systems around that money. That’s the legacy system we were dealing with. So yes, politics is part of it, but also incentives.

We took more of a Trojan horse approach. We built adaptive systems very early on so we could tell where students were missing things and support them in ways that were individualized.

So we started working not only with teachers but more directly with students as well – because I felt that maybe we had to disrupt the system from the learner side. But changing an entire system is hard. You have to move carefully. You take one step, stabilize, then take the next.

What has changed in the last three years is that AI finally gives us tools that make deeper change more possible. We now have a teacher assistant that shows educators where each student is and how they are likely to progress into the next concept. That kind of support can help create systemic change.

Ksenia:
Do you follow what teachers are asking for? What they want from AI?

Neeru:
Yes, very much. We work with teachers all the time, and we get many requests. But teachers are very different from one another. Some can imagine what’s possible and are ready to move. Others are afraid, tired, or not ready to take risks. So you have to work with many different profiles.

That’s one reason we’ve been around this long. Change in education takes a long-term view.

Ksenia:
At the systemic level.

Neeru:
Yes. There is still a lot of fear. But if people can see that the change is real and not scary, adoption can happen faster.

And of course there will always be bad actors. But that’s true of any technology. When we created content, we had domain experts and human review. We constantly checked quality – what the systems were saying, how students were interacting with them. That quality control still matters. Now automation can help, but responsibility still matters.

What Skills Matter Most Now?

Ksenia:
What skills do you think are essential in this AI world – for teachers and for kids?

Neeru:
Teachers need open minds. They have to be willing to think about what’s possible.

Every time a new tool appears, people panic. When calculators arrived, people said students would never learn math. When the internet arrived, people said students would just cheat. The same story continues.

Of course students still need to know how to think, compute, read, and reason. But some tasks don’t need to be done by hand forever. That doesn’t make the underlying understanding less important – it makes it more important.

Teachers need to understand how these systems work. Right now, many of them don’t. But they are smart. They will learn.

Students, meanwhile, still need literacy and numeracy. They need to write. They need to understand language. They need math. What’s interesting now is that language has become even more powerful. We once thought English majors might not be especially practical. Now prompt quality and language fluency are central to how people use AI.

We also need to help students connect their interests to deeper knowledge. If a child says, “I want to be a stage designer, so I don’t need math or science,” we can show them that they absolutely do – through lighting, structure, area, form, efficiency. Creativity and technical understanding are not opposites.

What EdTech Often Gets Wrong

Ksenia:
What do you think companies offering AI in education misunderstand about kids and teachers?

Neeru:
One big misconception is the belief that AI is enough by itself – that you can just ask a question and get what you need.

What systems like large language models do is generate the most likely fit probabilistically. But education has non-negotiables. Learning has structure. Development matters. Misconceptions matter. Prior knowledge matters. Current AI systems are not ready to handle all of that on their own.

That’s why I think every vertical needs to be treated seriously in its own terms. Education is not the same as medicine. These are different domains with different responsibilities. You cannot simply take a general model and assume that is enough.

Ksenia:
And when you say verticals, you mean domains like education and medicine?

Neeru:
Exactly. Those are different verticals built on top of these foundational models. And I don’t think we even fully know what AGI means yet, in practical terms.

Kids, Curiosity, and Learning by Doing

Ksenia:
One thing I keep thinking is that AI is becoming as important as reading and writing – a new literacy. And maybe we should also follow kids a little more, because they approach technology with imagination and curiosity.

Neeru:
I saw that even when my own children were in middle school. Kids have always had their own social graph – their own way of learning from each other, teaching each other, figuring out who knows what.

I spent more than 15 years deeply involved in education. Originally I was a molecular biologist doing research at Stanford on oncogenes. Then I became pregnant with my first child and decided I couldn’t stay around radioactivity. Later, after having four children, I started thinking seriously about education.

I found what I thought was the best school for my children – a very child-centered school. I got deeply involved. I spent about ten years in classrooms, watching what teachers and students were doing.

That was really the genesis of CK-12.

Ksenia:
And what does CK-12 stand for?

Neeru:
K-12 is obvious. But the “C” stands for every child – and also for the core C’s: creativity, critical thinking, collaboration, communication. Those are non-negotiables for deep learning.

Ksenia:
That’s interesting, because those skills were always important – but now they feel even more essential.

Neeru:
Exactly.

Why Free Is Hard in Education

Ksenia:
You said something interesting earlier – that because CK-12 was free, schools had trouble adopting it. That’s such a paradox. Free is so important, but it’s also hard to make work in education.

Neeru:
Yes. Most free products don’t last. We’ve lasted almost 20 years. But any system – nonprofit or for-profit – needs sustainable funding.

That’s why my husband Vinod and I have stood behind this. We fund it because we believe in funding every child.

Ksenia:
And how do children get access to it?

Neeru:
They can just join. It’s open and free. If they’re under 13, they need teacher or parent permission. That’s one thing that breaks my heart a little – because if you’re under 13, there are more barriers to independent learning on platforms like this.

Ksenia:
That is a real barrier. There’s also a big difference between open-source culture in software – where people help each other build things – and education, where openness still feels unusual.

Neeru:
Yes. When we started, there was more conversation around OER – Open Educational Resources. But not many survived.

Wikipedia was there, of course, but it wasn’t foundational for children. Most things online were not at the right level for kids. They didn’t account for prior knowledge, which is a major non-negotiable in learning. You have to figure out what the child already knows. Otherwise, if something is too hard or too easy, they lose interest.

That’s why cold start matters so much in learning systems.

Philanthropy, Access, and Human Potential

Ksenia:
If we go back to philanthropy, how do we make it more impactful in systems that are so hard to penetrate?

Neeru:
Education is one of those things we simply cannot ignore. It is a human rights issue. If the US – or any country – wants to remain strong, it has to make sure the next generation is equipped.

And I think this can happen fast, because in countries like India and China, children are hungry to learn. In many places, children understand that education is their path forward. If you are born into hardship, your family may not be able to help you study – but we can.

That is how you do it.

Ksenia:
My big hope is that AI can really accelerate that. Do you feel the same way?

Neeru:
Absolutely. AI lets us deeply trace where a child misunderstood something and help them recover faster. The current education system often expects a child to make up years of missing knowledge in one year. That’s nearly impossible without support.

That is why Flexi matters. It is there all the time.

Raising Grounded Children

Ksenia:
If I can ask something a little more personal – you really strike me as a grounded and modest person. With all the resources you have, how do you raise children to stay humble and grateful? What are your core values?

Neeru:
For me, the core values are about caring – for people, for animals, for the world around you. My children have inherited that.

I also don’t think privilege should be something you flaunt. We always encouraged our children not to chase the fanciest places or the most prestigious labels, but to explore. We’d go to places like the Amazon or the Galápagos – not for status, but for discovery.

When they were in middle school, we organized trips with other families and a teacher to South America, where the children lived in villages and helped with practical things – building toilets, helping where help was needed. We wanted them to see how most people actually live.

Those are the values that matter: help where you can, whether through money, mentorship, or simply showing up.

And I often think about how many educated women in the world have skills and talent that go underused. What if more of them helped mentor children around them? What if they became available to teach, guide, and support? That could go a very long way.

Ksenia:
I see some of that in my community too, but we live in a small town. It feels harder in big cities.

Neeru:
Maybe. But schools are still there. Parents can volunteer. In my children’s school, many parents brought their expertise into the classroom and helped teachers. You can do that. Show up.

Ksenia:
So it’s more about showing than telling.

Neeru:
Yes. Show up. All of us have to show up.

Intention, Not Just Attention

Neeru:
One more thing. In machine learning, people say attention is all you need. That’s what taught the machines. But for humans – for education, for society – attention is not enough.

We need intention. We need judgment.

Why am I learning this? Why does it matter? It cannot just be let AI do it. That’s what I would say.

Hope and Concern in the AI Age

Ksenia:
What is most exciting for you in this AI world right now?

Neeru:
Two things.

First, I’m going to become a grandmother for the first time.

Ksenia:
Congratulations!

Neeru:
Thank you. But I mean something larger too: humanity continues. Even though our time is limited, human beings continue. That matters.

Second, I really believe that if we use AI as augmented intelligence – something that helps us – we can do wonders. Don’t be afraid of it.

That excites me.

Ksenia:
And what are your concerns?

Neeru:
Fear is a real concern. And sometimes fear is grounded in reality. Human beings fight for dominance, for advantage, for shortcuts. There will always be people who try to game the system. We’ve seen that in every technological wave – from crypto scams to identity theft.

Those are real problems. But again, no technology arrives fully ready. Not even a baby comes ready-made. We have to stay involved and keep making systems better.

My concern is when a small number of people capture all the benefits and hoard them. That worries me.

Books That Shaped Neeru

Ksenia:
I always end with a question about books. What’s a book that significantly influenced you – either recently or from childhood?

Neeru:
There are several.

I learned to read English through Dr. Seuss and Archie comics. Later, The Hobbit influenced me a lot. And then Man’s Search for Meaning by Viktor Frankl had a very deep impact on me.

That book, with its reflections on concentration camps and how people kept their sanity in terrible circumstances, really stayed with me. It made me think about how we help children understand that life can be hard – but that people are also deeply adaptable and resilient.

Ksenia:
Yes. People are very adaptable, and we want them to adapt toward something better.

Neeru:
Exactly.

And I also love music. I love Hamilton. That line – I’m not throwing away my shot – really resonates with me. I’m still not giving away my shot. I’m going to keep taking it.

People say learning is lifelong, and I really believe that. I was around 50 when I went back to college for another degree. I was 52 when I started working on CK-12.

Ksenia:
What’s next for you to learn?

Neeru:
So much. One area I keep thinking about is economics – especially the economics of AI.

I’m not worried that AI will simply eliminate all work. Every major technological shift has opened new kinds of work while removing some forms of labor people no longer need to suffer through. AI will do some of that too. But people will have to reskill.

That’s exactly what we want our children to learn: learn broadly, learn quickly, keep adapting – but truly learn, and apply what you know.

This interview has been edited and condensed for clarity.

Further reading