Public sector organisations in Wales are being urged to take a more measured and purpose-led approach to AI adoption, as sector leaders warn that enthusiasm for new tools must not eclipse the basics of problem-solving, data governance and workforce readiness.
Speaking on the Business News Wales Government and Not for Profit podcast, senior figures from Perago, the Wales Institute of Digital Information and South Wales Fire and Rescue Service said that while support from the UK and Welsh Governments had accelerated interest in AI, organisations should resist pressure to adopt technologies before identifying the challenges they are trying to address.
Chris Elias, Communications Lead at Perago, said public bodies were increasingly being encouraged to explore AI, but warned against beginning with the technology rather than the problem.
Chris said:
“Don’t rush into AI as a product. First take some time to understand what you’re trying to achieve and where AI, in amongst other technologies, could play a part.”
He argued that many organisations still felt obliged to “do something with AI” without first being clear on the purpose, and that this made it difficult to measure success later.
He said some public services had already moved prematurely, and pointed to examples discussed at a recent event in London which suggested only a small proportion of UK local authorities had seen any return on investment from AI tools. Chris said this often reflected projects that began without clearly defined objectives, making it harder to demonstrate impact or value for money. He also highlighted the importance of creating a culture where staff can experiment safely and raise concerns about risks such as job security, data handling or trust in technology.
Tim Bashford, Research Lead at the Wales Institute of Digital Information at University of Wales Trinity St David, said it was too simplistic to frame AI adoption as a divide between public and private sectors. He said larger organisations across both sectors tended to be more digitally mature and therefore further ahead, while smaller organisations, including many SMEs, were “really underprepared.” Tim said Wales had developed comparatively mature governance for AI, but adoption remained uneven because of legacy systems and varying levels of digital capacity.
He added that higher education had a major role to play beyond specialist research. Tim said:
“We can lead and facilitate training. There is a need for practical CPD that gives people the skills they need to use AI responsibly in the workplace.”
He argued that staff across public services would increasingly need training in areas such as prompt literacy, ethical use and understanding when not to rely on automated outputs. Universities, he said, were already beginning to develop accessible training to help organisations adopt AI safely.
For South Wales Fire and Rescue Service, the focus is on building firm foundations before deploying new technology. Head of Communications and Engagement Rhian Moore said the Service was at the start of its journey but was intentionally taking a structured approach.
Rhian explained that digital transformation work was looking first at essential systems, infrastructure and user needs so that AI could be assessed as part of a wider effort to improve efficiency, productivity and public value. She said the organisation had established an AI working group with representation from across the business to consider issues such as cybersecurity, confidentiality, ethics and transparency.
Rhian also pointed to the importance of training in building confidence. She said the Service was already working with early adopters internally and exploring practical benefits, from streamlining processes to using data more effectively for prevention work. She emphasised that in a safety-critical service there were limits to what AI could replace, but significant potential for it to support operational preparation and help identify vulnerable communities.
Hear more in the Business News Wales Government and Not for Profit podcast episode Ready for AI. Listen here.
