Artificial intelligence is increasingly being used to support the delivery of public services. From GOV.UK Chat to smarter procurement recommendations, AI is helping citizens and government teams navigate complex information and make better decisions. To ensure these systems are safe, fair and effective, the Cross-Government Testing Community has developed the AI Testing Framework for the public sector. The framework provides: ➕ principles for testing and evaluating AI systems ➕ key quality attributes to check such as fairness, safety and reliability ➕ ongoing assurance throughout the process ➕ flexible testing methods for different types of AI If you would like to learn more, check out our latest blog here https://lnkd.in/ezXfQiiQ #DigitalGovernment #AIInGovernment #GDS
Adam Price
3w
I see there is mention of efficiency and an ethical review, but it’s hard to see how we can do that effectively when even the heads at MIT technology review struggled to model energy use due to the lack of disclosed information - https://www.technologyreview.com/ 2025/05/20/1116327/ai-energy-usage-climate-footprint-big-tech/ That seems like something we might need to acknowledge in evaluating and assuring AI projects, and potentially include a steer earlier in the process to explore less energy-hungry, custom-built AI models that don’t require LLMs and the vast amount of parameters we see (estimated, given the lack of transparency in the area!) with models like ChatGPT and other large AI tools.
Johannes Neumeier & Dr. Chris Richter
James Kavanagh
1mo
Really useful framework, thanks to the team who put this together
This is exactly the kind of structured approach public sector AI needs. Principles, quality attributes, and ongoing assurance are essential to building systems that are safe, fair, and reliable. Flexible testing methods are particularly important given the variety of AI applications in government from chatbots to procurement decision support. Rigorous evaluation frameworks like this help ensure AI delivers real value without compromising trust or accountability.
Attul S.
1mo
Jari-Pekka Sova, please see attached.
UK Data and AI Lead at Netcompany
1moAnahita S.