Building trustworthy technology: How AI and data can improve government | Computer Weekly
Briefly

Building trustworthy technology: How AI and data can improve government | Computer Weekly
"The rise of artificial intelligence (AI) has been met with both excitement and anxiety - a blend of optimism about what could be achieved and unease about what might go wrong For every bold promise of faster discovery, fairer systems and more efficient public services, there are stories of bias, opacity and overreach. We've seen AI tools in recruitment perpetuate inequality, and algorithmic systems in welfare and policing deepen discrimination rather than dismantle it."
"At the Open Data Institute (ODI), our starting point is simple: data and AI should be treated as public infrastructure. Like roads, railways or the NHS, data systems need to be built, maintained and governed in the public interest. When data is fragmented, outdated or locked away, everything built on it becomes weaker. But when it's accessible, interoperable and stewarded well, it can drive innovation, efficiency and trust. You can see that principle in practice through OpenActive, the open data standards for physical activity."
Artificial intelligence brings both promise and risk, offering faster discovery, fairer systems and more efficient public services alongside bias, opacity and overreach. AI tools have perpetuated inequality in recruitment and deepened discrimination in welfare and policing. Data and AI should be treated as public infrastructure, requiring building, maintenance and governance in the public interest. Accessible, interoperable, well-stewarded data enables innovation, efficiency and trust. OpenActive demonstrates open data standards connecting leisure operators, local authorities and platforms to create millions of discoverable opportunities. Integrating open data into health systems could enable preventative health measures such as GP exercise prescriptions and support long-term health plans.
Read at ComputerWeekly.com
Unable to calculate read time
[
|
]