TRANSPARENCY
We use AI personas to question our designs and to find problems we might miss.
The board is eight AI personas. Each one is an AI model given a single specialist point of view and told to look for problems rather than agree. We give them the material in question, such as code, the instructions we give the AI, or a design, and ask them a question.
The advisors are AI, not people, and their reviews are not professional advice.
Teaching and language
Checks how the app teaches and how AI marks answers, and whether the marking is fair to every learner.
Design and accessibility
Checks whether every screen is easy to use and accessible, on a phone as well as a computer.
Security and data protection
Checks how accounts and children's data are protected, and flags questions that need a lawyer.
AI reliability
Checks which AI models we use, how they can fail and what they cost.
Product and rewards
Checks whether points, credits and games help students learn, rather than just keep them clicking.
Testing and data
Checks that data is saved correctly, that scheduled jobs work and that our tests cover what matters.
Books and reading packs
Checks which books we use, the words they offer and how the reading packs are put together.
Attacks and misuse
Looks for ways someone could break in, cheat the credits or trick the AI, and for risks to children from other users.
AI advisors are no substitute for real experience. If you work in education, child safety, accessibility, language or app development, or simply care about how children learn, we'd love to hear from you.