THE BIG DEBATE
The systems businesses use can produce plausible material that is incomplete, biased or simply wrong. In finance, recruitment, health, publishing and other regulated or high-risk areas, an efficient process is not necessarily a safe or defensible one.
Lumsden-Cook points to warnings issued in 2023 about AI-generated mushroom identification guides offered for sale online. Where poisonous and edible species can closely resemble one another, an unchecked error is not a minor editorial flaw. His argument is that verification, credibility and accountability are part of the value a responsible publisher provides.
The same principle applies across industry. Somebody must own the outcome.
The council’ s business case acknowledges the challenge. It highlights procurement, intellectual property, cyber security, data protection, transparency and accountability as areas requiring governance, assurance and legal input.
Davies is enthusiastic about the possibilities but equally candid about the absence of settled rules.
“ I’ m on both,” she says.“ It’ s definitely exciting. The opportunities are abundant and infinite. I’ m also on the, you know, God, we’ ve just connected Claude up to our SharePoint, and what the hell does it know, and what can it do with that?
“ Having some guide ropes around that is important and I just don’ t think we’ ve got enough. So I’ m scared by it as well.”
Different employers are taking markedly different approaches, she adds. Some encourage staff to experiment with multiple tools, others approve a single platform and some prohibit their use.
It is, she says,“ a bit of the wild west out there”.
There is another strategic risk for businesses whose competitive advantage lies in proprietary knowledge. ProtectaPet’ s bespoke configurator contains expertise developed within the company. Making more of that knowledge accessible through an AI-driven customer tool could improve the service and create an early advantage, but it could also expose valuable learning to competitors.
That tension- between extracting value from data and protecting it- is likely to become one of the defining management questions of the next few years.
A widening gap?
AI is often described as a democratising technology because powerful tools can be accessed cheaply, sometimes free of charge. A small company can perform work once available only to a much larger organisation.
But today’ s prices may not last. The systems are costly to develop and operate, and providers will eventually need sustainable returns. Subscription charges, usage limits and paid-for tiers could rise as businesses become more dependent on the tools. Advertising or sponsored answers could also influence the information users receive.
If the best models, infrastructure and training become available mainly to organisations able to pay, AI could reinforce the advantage of scale instead of reducing it. A similar divide could emerge between workers equipped to use and challenge the technology and those whose roles are easiest to automate.
That is why the council’ s focus on shared infrastructure, business support and inclusion matters. Its preferred approach is to build on existing organisations rather than create an entirely separate programme, while coordinating local authorities, universities, training providers and employers. Funding, however, has not yet been committed, and the report says further work is required on costs, staffing and sources of investment.
The £ 2 billion opportunity is therefore an estimate, not a guaranteed dividend. Reaching it would require businesses to adopt AI effectively, workers to develop the right skills, infrastructure to keep pace and safeguards to earn public trust.
Test, learn … and keep thinking
For an individual business, the most sensible response may sit between
“ Having some guide ropes around that is important and I just don’ t think we’ ve got enough. So I’ m scared by it as well.”
resistance and unquestioning adoption.
Davies argues that each company must identify where AI genuinely saves time and where it might weaken decision-making.
“ They’ ve got to learn where AI gives them efficiencies and time savings and is helpful, and conversely, where AI is actually undermining their ability to make really meaningful decisions, make the right decisions,” she says.
That means starting with a real problem rather than adopting technology for its own sake; testing on a manageable scale; protecting sensitive information; measuring the result; involving employees; and keeping a human accountable for important decisions.
It also means resisting the assumption that technological progress is automatically social or economic progress. AI may create new industries and more valuable work, but transitions produce winners and losers unless institutions and employers actively shape them.
Staffordshire’ s ambition is deliberately bold. Becoming the intelligent heart of the country will depend on more than the number of businesses using AI. It will be measured by whether the county can translate faster processes into stronger companies, better jobs and wider prosperity— without allowing short-term efficiency to hollow out the knowledge, trust and human relationships on which sustainable businesses depend.
The two perspectives that follow explore different parts of that challenge. They are not exhaustive cases for embracing or rejecting AI. They are takes from business leaders already dealing with its consequences: one looking at the time and capacity it can release for small firms, the other at what may be lost when technology removes the work through which human expertise is formed.
26 ISSUE 7