Building AI you can trust in education
Education is a setting where the stakes are high and the users are young. We set out the principles that govern how we build responsible AI, from privacy and transparency to the choice never to optimise for engagement.
Building artificial intelligence for education carries a particular responsibility. The people using our tools are often children, the decisions involved shape futures, and the trust placed in a learning platform by a parent or a school is not lightly given. We take that seriously, and it shapes how we build.
The first principle is privacy. Student data is among the most sensitive data there is. We collect only what is needed to support learning, we are clear about what we hold and why, and we do not treat a child's learning history as a commercial asset to be traded or mined.
The second is transparency. When our tools reach a conclusion, whether marking a piece of work or identifying a misconception, they should be able to explain how. A student, teacher, or parent deserves to understand why the system said what it said, rather than being asked to trust a black box.
The third is restraint. Many digital products are designed to maximise the time users spend with them. We deliberately do not build for engagement. A learning tool that keeps a child on the screen longer is not necessarily helping them learn; often the opposite. Success for us is a student who understands something and moves on, not one who stays.
The fourth is humility about limits. AI systems make mistakes, and a system that presents every output with total confidence is dangerous in a classroom. We design our tools to acknowledge uncertainty and to keep a human in the loop for decisions that matter.
These principles are not a finished charter. Responsible AI in education is a practice that has to be revisited as the technology and our understanding change. But they are the standard we hold ourselves to, and the standard we invite others to hold us to.
