Eating Light

The concept is called “children pop-up on-site lessons” and they can arise anytime, anywhere. They are incredibly inspiring. They are never planned and are always guided by children. What you can learn from them depends entirely on your ability to observe and listen. The opportunity presents itself right there, in that precise instant. It is up to you to absorb everything that is generously and selflessly offered—children's pop-up on-site classes are never paid for.

I can assure you that I have attended many classes throughout my life. None are like this. None have the power to transform me as profoundly as this one.

My latest children's pop-up on-site class happened just a couple of days ago on a flight, where a three- or four-year-old girl traveling with her parents gave all of us a masterclass with profound implications for how museum and education professionals understand our work. It challenged how we create content and learning environments within our respective organizations. Even on the way we conceive AI. Serendipity.

The girl and her parents were seated six rows ahead of me. We were about to take off, and at one point, the girl—standing on her mother’s lap, securely held—felt a burst of cool air on her face and looked up. She noticed something that caught her attention: the mysterious opening of the air vent and a small light indicator. Immediately, she started touching and pressing them over and over again—a process she continued for the entire flight time.

And then, extraordinary things began to unfold:

• The girl transformed something as seemingly trivial and mundane as a cabin call button into a unique learning experience. She needed no resources, no classroom—nothing at all. Just what she found around her, which she effortlessly turned into something interesting and educational.
• Her main learning tool was play. To understand what was capturing her attention, she played. She played using different strategies: touching, moving, shouting, laughing, observing, continuing to explore, comparing, trying to make noise with the buttons… Everything was part of the game. Play was her way of approaching the unknown, dismantling it, and making it her own.
• Her connection to the present moment was absolute. Fifty-five minutes focused on a single point. Nothing else existed. Nothing distracted her from what she was experiencing at that moment. That’s what made it unique—her ability to ignore everything that was not the glowing button. What a powerful lesson for our societies, where our attention is constantly extracted and monetized.
• She engaged all her senses in the discovery process. She didn’t just see and hear; she touched, smelled, and experienced her entire sensory system at once. It was fascinating, for instance, how she used her sense of hearing: after realizing that pressing the button made a sound, she kept doing it, triggering laughter among the passengers and even from the flight attendant, who let her continue. Humor is the reverse side of getting skills. But her approach to taste was revolutionary. Could she eat the light? Why not? She saw something she liked in the same way she would see a strawberry or a jar of jam—so she put it in her mouth. She reached for the light, brought her hand to her lips, and pretended to chew and swallow. She was eating light, right in front of us! She invited us to do the same by teacher how to do it. What museum installation could work in such an incredibly out-of-the-box way?
• The environment played a crucial role. Her mother let her act freely but safely. From time to time, the girl would pause for a moment and celebrate her discovery—by using her mother’s head as a drum, sometimes a table, or as a landing planet surface. The mother would burst into laughter every time. It was pure joy, expressed through shared body language, connecting with her parents and with all of us.

We arrived at our destination, and the girl was still pressing the buttons. She could have gone on for hours. A whole new world had just opened up to her, and she was only beginning to grasp it. That made her happy. When they left the plane, the setting changed, but her approach remained exactly the same. Now, the moving walkways in the airport became her/our next big opportunity—testing her balance as she moved forward.

Children's pop-up on-site lessons are not just extraordinary opportunities for unlearning through learning. They also have deep echoes in the way we are creating technology. There is a direct connection between what this girl did on the plane and what DeepSeek has recently accomplished. But how? Where’s the link?

The founders of DeepSeek, a Chinese start-up backed by $5.6 million and powered by 2,048 Nvidia H800 chips, have triggered their Sputnik moment by reversing the conventional approach used by companies like OpenAI or Meta when training their LLMs (Large Language Models).

Typically, AI models are first trained through supervised learning, where they are fed labeled data that humans have pre-tagged with correct responses. Only afterward do they apply reinforcement learning, which works through trial and error, using unlabeled data and a system of rewards and penalties.

What was the girl on the plane doing? Reinforcement learning.

DeepSeek flipped the process—starting with reinforcement learning first. This means less training time, which translates into less energy consumption, and the ability to use older, cheaper microchips instead of state-of-the-art ones.

The girl on my flight was, in front of our very eyes and with pure intuition, demonstrating a prodigious exercise in reinforcement learning—based on triumphant discoveries, which she shared in real-time, in open-source mode, just like DeepSeek.

A pure surge of endorphins, showing all of us the path to follow—one glowing button at a time.

Magic can appear unexpectedly, hidden in the folds of reality. On a short flight -like this time-, at a bus stop or while waiting at the supermarket checkout. Are we able to detect it? Please, take a look at your inner child connection state.