At the Open Knowledge Foundation (OKFN), we have spent over two decades championing open technologies and standards in the public interest. Our latest work under the AI Learning Labs initiative asks a pressing question: how can we use AI to make public data more accessible without sacrificing accuracy or trust?
Public data is essential for accountability, participation, and informed decision-making. Yet too often, citizens, researchers, and journalists must navigate complex portals and master technical queries simply to extract answers. That is why we created a field guide to share what we have learned about connecting AI to government data responsibly.
We are publishing the results today: The Tech We Want | AI Learning Labs — Connecting AI to Government Data
The guide is part of The Tech We Want series, our ongoing conversation about creating digital tools that are simple, long-lasting, and genuinely useful, as a counterpoint to the overengineered stacks of Big Tech companies.
Key content of the guide includes:
🧑🏽🍳 A pilot ‘recipe’ — the team, infrastructure, and timeline required to help you replicate our approach, from aligning on goals to testing and reflection.
📋 6 steps to a data provenance framework for AI agents, ensuring every answer has a clear trail back to the source record.
🤗 Practical recommendations for humans and machines, covering everything from separating facts from interpretation to validating arithmetic calculations.
⏸️ A reflective section acknowledging the limits of the experimentation, including the ‘blank chat’ problem.
The main message is simple: clear provenance helps improve AI-assisted access to public data, although there is no one-size-fits-all approach. The tools we use to publish and query open data must evolve to match how people and systems ask questions now – but you need to know what they are!
We would love to know: how have you approached data provenance in your public interest AI projects? Share your reflections with us at info@okfn.org and join the conversation on the Open Knowledge Forum.
Acknowledgements
Special thanks go to the teams from the Brazilian Office of the Comptroller General (CGU) and the Agency for the Electronic Government and for an Information Society in Uruguay (AGESIC) for their partnership in the pilots.
We are thankful for the support of the Patrick J. McGovern Foundation. Learn more about its funding programmes here.
See the previous guides in this series…
