Interview reflection: Mandar Kulkarni, THE Labs
After the interview, each student should write their own short reflection on what they heard, learned and want to remember.
Skye Shi and I interviewed Mandar Kulkarni on October 22. Mandar is the co-founder and CEO of THE Labs in Pune, a nonprofit that teaches AI and the humanities to children in grades 6–8 from underserved communities. This year it reaches 4,000 children in government, government-aided and tribal schools across four districts of Maharashtra. About 11 facilitators each teach groups of 20 children, an hour a day.
AI as a subject
THE Labs teaches children how AI works so they can question it. How to use it gets little attention. Mandar put the difference this way: “if it is a tool, I’m going to use it, but if it is a system, probably system is going to use me as well.” Children build projects in groups of five on custom tools that keep them away from public AI products, and they debate cases like copyright disputes over AI art. In one project, children carried out a study, designed by the team, that measured caste bias in ChatGPT. The assessments ask children to work backwards: given an AI’s output, what might have gone wrong in its training data?
How the plan changed
In December 2022, Mandar and his co-founder Mayura planned to open a school. They had finalized the land and looked for an architect. ChatGPT had come out a few days earlier, and they decided their idea, teaching technology and the humanities together, should “reach many more children” than one school could. By October 2023 it had become an after-school program.
The curriculum changed in a similar way. The team wrote about 200 hours of material and piloted it with a group chosen on purpose for its range of reading and math levels, so they could see how those levels related to how children did in the program. Mandar told us that 50 to 60 percent of sixth graders in India can’t read text at a second-grade level. After the pilot, they carved out a 100-hour core that, in his words, “every child can cope with,” and that became the AI Citizenship Program. The rest of the material is still being piloted, possibly as a deeper program for some of the children who finish the core. THE Labs now runs two tracks: the core at scale in schools, and depth at centres of excellence meant as demonstration sites. The first centre is in Pune, and they plan six. When I asked whether the centres were a return to the school they first planned, he said, “kind of.”
What it means for my design
This is the problem my own course redesign is trying to solve. In Blog #3 I wrote that I’ve sized my required path by judgment. Mandar sized his core by piloting with a deliberately mixed group and seeing what all of them could manage. I’d like to do something closer to that, and look at how students with different backgrounds actually do on each unit before deciding what everyone has to master.
I also put to him an argument I hear often in New York, where the mayor has placed a moratorium on AI in K–12: AI is easy to pick up, so children can learn it later and be protected from it while they’re young. His answer was that using AI is the easy part, because the tools “are built to be user-friendly.” THE Labs teaches “how it works and how to challenge it,” and privileged children pick up that kind of understanding at home, from parents who work with these technologies. The parents of his students are delivery workers and household helpers. He also sees the program build confidence at an age when, without it, many of these children leave school after 10th or 12th grade.
What I want to remember
At the end I asked what one piece of advice he’d give someone trying to make a change in education. He said, “I would say be with children, as much as possible.” He spent all of his first year with children, and as CEO he still observes classrooms and teaches sessions himself. He added a caution about student agency: children won’t always make the right decision, and an educator has to keep student voice in balance. I told him it was the best advice I’d heard in a long time.
My redesign started from what students told me when I asked why they skipped optional work. The condition I said I’d bet against in Blog #3, whether students can tell when they need more practice, is one I can only answer by being with them while they work: sitting with beginners in office hours and study groups as they go back over early material, or don’t.