Co-founder, Algorithm of Learning. Princeton 2015
Exploring how humans learn — and think — in the age of AI.
Building @gakkuai, the AI learning companion.
Expressing ourselves matters more, not less, in the age of AI. It helps us figure out what we think—and who we are. Watch my first podcast interview—hosted by Raku, our AI learning companion 😄
We’ve raised $51 million to rethink how kids learn with AI. Our bet: conversational learning.
Language isn’t just how we communicate. It’s how we think. That’s the idea behind Gakku, our AI learning coach. The learning process itself is the product—not something to automate away.
Does expression matter more in the age of AI? This time, the AI is asking.
Watch Raku, our AI learning companion, interview co-founder Annie Chen @anniechen419 , who created him.
I came to the U.S. at 16 for high school, then stayed for college. The world felt more open back then. Different accents, backgrounds, and ways of thinking weren’t things you had to explain away. They made the room smarter.
I know it's a different world now. Maybe I’m nostalgic, but I miss that openness. I still believe in it. In meritocracy. I still want what you build and contribute to matter more than where you’re from.
Watching the pile-on around Jensen lately, I feel like it’s gone too far. Of course he’s not above criticism. But there should be room to hold someone accountable without reducing everything about them to politics and stereotypes.
And honestly, I’m tired of scrolling through X and seeing AI memes about him turn into racist, sexist jokes about Chinese and other Asian women.
@allanlundhansen@fchollet Thanks! and fair point — we don’t mean that all human thought is linguistic. More precisely, language is one of the ways we shape, examine, and refine our thinking, especially when we’re reasoning with others or trying to make an idea explicit.
@TheElliottSmith@fchollet great question. the key is to decide when to encourage/explain vs to challenge/push back. So if a kid already makes efforts and stays stuck for a while, gakku will explain the answer and the why behind it. and of course the kid can ask questions along the way.
Thank you, François — this means a lot!This is exactly the distinction we care about: the learning process isn’t friction to eliminate. It’s where thinking gets formed. The opportunity for AI is to make that process more personalized, interactive, and expansive—without taking agency away from the learner.
@AtoAUnity@gakkuai thanks Arthur! yes, and actually when to push back vs. just encourage also depends on who the user is, it should be personalized as well.
Thanks for sharing this, Aran! Learning through dialogue goes back to Socrates and Confucius. AI could now give everyone access to a powerful thinking partner. The question I keep coming back to is how we really keep the human at the center— how we can expand thinking and explore who we want to become.
Expressing ourselves matters more, not less, in the age of AI. It helps us figure out what we think—and who we are. Watch my first podcast interview—hosted by Raku, our AI learning companion 😄
We’ve raised $51 million to rethink how kids learn with AI. Our bet: conversational learning.
Language isn’t just how we communicate. It’s how we think. That’s the idea behind Gakku, our AI learning coach. The learning process itself is the product—not something to automate away.
Does expression matter more in the age of AI? This time, the AI is asking.
Watch Raku, our AI learning companion, interview co-founder Annie Chen @anniechen419 , who created him.
@omarsar0 Thanks, Elvis! Exactly, explaining and communicating aren’t just ways to share thoughts we’ve already formed—they often spark new ideas and make the thinking process itself visible.
@andrew_n_carr Thanks a lot for sharing this, Andrew!Drawing is such a wonderful way for your son to express himself. I’d love for learning to build on that desire to create, rather than feel like something separate.
@omarsar0@gakkuai Thanks, Elvis! For us, the key is building an experience-led product—intellectually challenging and inspiring, but also genuinely fun and engaging (and that’s why real-time multimodal interactions is so important to us). That’s the balance we’re trying to get right with Gakku!
@kimmonismus Sometimes the hardest part isn’t getting AI to understand what we want—it’s figuring out what we want in the first place. Thanks @kimmonismus for capturing this so well!
Thanks for trying Gakku with your 6yo — really curious to hear how it goes!
“How much help to give” is probably the hardest question. The right amount depends on the learner and the task: enough to expand their thinking, but not so much that AI takes ownership of it.
That balance is where a lot of the learning happens.
Love this framing from @LinusEkenstam: the interface becomes part of the thinking. The real design challenge is knowing when AI should guide, expand the space of possibilities, or step back and let the learner own the next move.
The biggest mistake in AI education is automating the part where the child learns.
The questions. The wrong turns. The “wait, let me try again.”
That’s the part gakku is built around.
@gakkuai helps kids aged 6-16 practice writing, speaking and thinking through live conversations with Raku, its AI learning companion.
What interests me from a product-design perspective is what happens when the child changes their mind. Having small kids myself, I know this happens a lot.
A learner can interrupt, revise an idea or make a choice on screen. Raku has to adapt while keeping the conversation, visuals and mission progress in sync.
The interface becomes part of the thinking. Ideas take shape on screen, and the child’s choices shape what happens next.
That’s a much more interesting design problem than generating a good reply. The system has to follow the learner’s intent, remember what came before and decide when to ask, act or wait.
For me, the design challenge is knowing how much help to give. Too little and the child gets stuck. Too much and the system takes over the decisions they need to practice making.
The child still has to explain, choose and revise.
The interesting bet here: the time spent thinking is the value of the product.
When AI can generate almost any answer, helping someone develop their own becomes a lot more interesting.
Currently trying this with my 6yo at home, will report back on how it goes.
Thanks for trying Gakku with your 6yo — really curious to hear how it goes!And thank you for all the product feedbacks!
“How much help to give” is probably the hardest question. I don’t think the goal is to minimize AI help, but to give the right kind of help at the right moment.
And that depends on the learner: who they are, what they’re trying to do, how they like to learn, and how hard the task is.
Ask when a question can unlock something. Hint when they’re stuck. Introduce a new possibility when it can expand their thinking. Step back when the learner needs to make the choice.
That balance is where a lot of the learning happens.
@dr_cintas Thank you, Alvaro. You’ve captured the bet behind Gakku: the next generation of consumer AI won’t just generate content—it will participate in experiences as they unfold.
I agree that kids need to do the hard work. But play and serious thinking aren’t necessarily opposites.
Putting a game-like skin on rote exercises is very different from designing an experience that requires thought. Games like Civilization involve planning, weighing trade-offs, and persisting through setbacks. Multiplayer games can also require real collaboration.
At Gakku, we’re aiming for immersive interactions that invite learners to reason, create, connect ideas to their own lives, and use their imagination. The thinking is part of the play—not something you finish to collect a reward.
AI opens up new possibilities here. Its ability to interpret long, open-ended conversations, combined with memory across sessions, can help us build a richer, more continuous understanding of a learner’s ideas and learning behaviors—not just whether they selected the right answer. Real-time voice, visuals, and interactive activities also let learners talk through ideas while exploring and creating.
Engagement alone isn’t proof of learning. But the choice isn’t between entertainment and doing the work. We’re trying to make meaningful thinking, practice, and creation the engaging part.
I agree that kids need to do the hard work. But play and serious thinking aren’t necessarily opposites.
Putting a game-like skin on rote exercises is very different from designing an experience that requires thought. Games like Civilization involve planning, weighing trade-offs, and persisting through setbacks. Multiplayer games can also require real collaboration.
At Gakku, we’re aiming for immersive interactions that invite learners to reason, create, connect ideas to their own lives, and use their imagination. The thinking is part of the play—not something you finish to collect a reward.
AI opens up new possibilities here. Its ability to interpret long, open-ended conversations, combined with memory across sessions, can help us build a richer, more continuous understanding of a learner’s ideas and learning behaviors—not just whether they selected the right answer. Real-time voice, visuals, and interactive activities also let learners talk through ideas while exploring and creating.
Engagement alone isn’t proof of learning. But the choice isn’t between entertainment and doing the work. We’re trying to make meaningful thinking, practice, and creation the engaging part.