UNESCO puts ethics first as educators rethink how AI should be built for learning

UNESCO reports that it has trained more than 35,000 civil servants from 190 countries in designing public services that integrate AI and uphold ethical governance.

Poor learning outcomes, a growing gap between the skills people acquire and the jobs available to them, and questions about the purpose of higher education are pushing education leaders to examine what artificial intelligence can offer. At Digital Learning Week 2026, UNESCO brought together policymakers, educators and innovators from more than 20 countries for hands-on training in designing AI-based education products and services. Participants were asked to examine the needs of the people who would use or be affected by these systems, placing ethical decisions at the beginning of the design process.

Understanding the problem before choosing AI

The training challenged a familiar approach to adopting AI, in which an organization chooses a technology and then searches for something to do with it. Participants began by mapping stakeholder groups and identifying the problems those groups face, creating a clearer picture of where support was needed. Their discussions covered poor learning outcomes in elementary schools, mismatches between skills and employment, the role of higher education in the 21st century, and concerns that AI could weaken teachers' authority or undermine the value placed on their expertise.

Beginning with these concerns gave participants room to consider whether an AI system would actually help, how it could complement changes that do not depend on technology, and which problems might need a different response altogether. For policymakers, this approach creates an opportunity to understand risks before making adoption decisions and to plan measures that reduce potential harm. It also keeps attention on the educational challenge being addressed, choosing technology part of a considered response to people's needs.

Building ethical safeguards into the design

Participants used a simplified version of UNESCO's Ethical Impact Assessment methodology to examine proposed AI systems against the problems they had identified. Grounded in UNESCO's Recommendation on the Ethics of Artificial Intelligence, the exercise required them to describe each system in detail, consider its ethical implications and map out procedures for reducing risks before deployment. Working through those questions during design helped connect broad ethical principles with practical decisions about how an education product or service would operate and affect the people around it.

The exercise treated ethics as a design responsibility that shapes a system from its earliest stages, giving participants a chance to address concerns before they become embedded in a finished product. Reviewing a proposed system before it is built makes it possible to reconsider choices and develop safeguards as part of the original design. This approach moves ethical assessment beyond a final compliance check and reduces the need to add protections only after a system has been introduced and problems have emerged.

Preparing educators and governments for responsible AI

Defining the problem before selecting technology and assessing ethical implications before deployment are central to UNESCO's wider work on AI, education and the public sector. The organization supports governments in developing AI governance that serves people, respects human rights and includes the communities affected by its use. Its competency frameworks for civil servants, students and teachers provide guidance tailored to these groups, recognizing that responsible AI use requires understanding and informed judgment from public officials, educators and learners themselves.

UNESCO reports that it has trained more than 35,000 civil servants from 190 countries in designing public services that integrate AI and uphold ethical governance. The Digital Learning Week workshop brought that broader approach into education through practical exercises, giving participants experience in questioning a proposed use of AI, identifying possible harms and planning responses before deployment. The training placed educational needs, teachers' expertise and the interests of affected communities at the centre of decisions about what to build and how it should work.