Lunchbox AI is the next phase of Ryelands AI Lab. In that project we worked with a Lancaster primary school to co-design and deliver six lessons introducing children to Generative AI through text-to-image generation. It worked really well, but it only reached two classrooms in one school, and much of its success came from having the research team in the room. Lunchbox AI is our attempt to take what we learned and make it work at a much larger scale, without us there.
Despite all the talk about AI literacy, there are very few hands-on tools designed for younger learners. Primary-aged children tend to get left out of GenAI education because of worries about risk, complexity and infrastructure. Lunchbox AI is a web-based platform for exploring GenAI in the classroom. Like Ryelands AI Lab it’s built on a constructionist approach, the idea that children learn most deeply by making and doing. The aim is to build AI literacy in both pupils and teachers, covering not just how to use AI but how to think about it.
The design came out of interviews with teachers and educators, where we asked whether the approach from the trial could reach hundreds or thousands of schools. From those conversations we drew up a set of design heuristics, then set about turning them into a working system. The image above shows screenshots of the working system. The teacher dashboard is where teachers open and close sessions and keep track of their class, and the learning space is where the class watches lesson videos and generates images together. We have written up that design process in a pictorial, which we’ll share once it’s published.
We’re now starting to trial the new system. The first step is a series of playtest workshops where colleagues get hands-on with the platform and then tell us what works and what doesn’t. After that, Lunchbox AI will become a living lab for testing our constructionist approach in schools at scale. I’ll post more as the trials get going.
