IBM: AI adoption stalls as UK businesses lack clear strategy
UK businesses are adopting AI at a pace, but many are still struggling to convert that momentum into real, meaningful financial returns.
According to IBM’s Race for ROI study, 66 per cent of British enterprises report significant productivity improvements from AI, with 63 per cent of senior leaders citing noticeable efficiency gains post-adoption.
There seems to be a gap between early success and wider transformation; however, as the same research shows, 62 per cent of organisations are still failing to unlock AI’s full potential. And that disconnect is playing out across boardrooms and business units alike.
“It’s not a technology problem, it’s an organisational change problem,” Sebastian Weir, executive partner at IBM, told City PM. “You can’t give a workforce access to a co-pilot and expect them to just get on board with it.”
AI adoption vs scale
Over three-quarters of UK businesses are using AI tools in some capacity, while IBM’s findings show companies are already benefiting from time savings and improved efficiency.
Those gains are translating into tangible changes in how work is done, with IBM reporting that AI is freeing up time for higher-value tasks, including innovation (41 per cent) and creative work (41 per cent).
Yet, a study by consultancy Studio Graphene found that progress beyond those early signs of gain is uneven.
Its report showed that 31 per cent of UK firms using AI have seen any positive return on investment, while fewer than half can clearly define what success looks like.
Similarly, research across global enterprises suggests that fewer than one in 10 companies is achieving a large-scale financial impact from AI.
“The friction we’re seeing is how you start to change behaviours, from leadership KPIs right through to changing the behaviours of an organisation,” Weir said.
Without those changes, AI tends to remain confined to isolated use cases rather than driving broader business transformation.
Unrealistic expectations
A key barrier is the level of expectation that AI will deliver immediate financial returns.
IBM’s research proves that while over a quarter of UK firms are already seeing cost savings, a further 34 per cent expect returns within the next year, as benefits often take time to materialise.
“Expecting sizable immediate return is unrealistic”, Weir argues. “It’s not that the benefit doesn’t exist, it’s about being realistic about the horizon.”
Much of AI’s early value is indirect, and efficiency gains, such as reduced time spent on routine tasks, are difficult to measure in traditional, quantifiable, financial terms.
“If I save three minutes from a phone call, it’s very difficult to cash that saving,” he added. “But it does make me more effective.”
This creates tension at the board level, where investment decisions are often tied to short-term performance metrics rather than longer-term productivity gains.
Skills and model bottlenecks
Alongside measurement challenges, workforce readiness remains a significant constraint.
IBM’s study found 67 per cent of UK business leaders cite internal resistance and cultural barriers as key obstacles to AI adoption.
And at the same time, only 38 per cent of organisations are actually prioritising AI upskilling across their workforce.
What’s more, a YouGov survey commissioned by The Access Group found that this gap is reflected in how employees use technology: 70 per cent of UK employees are experimenting with AI tools, while only 19 per cent have received formal training.
“There was quite a lot of organisations that released AI guidance three years ago saying, ‘Don’t do this.’ Very few have refreshed that to say, “Okay, now you can,” Weir added.
“Everyone needs critical thinking skills to understand what is right, what is wrong, what do they trust”.
The speed of advances in AI models is also making it harder for businesses to keep up, with new systems and updates released frequently.
While this has fuelled investment, it has also made it harder for organisations to stay focused on outcomes.
“It’s incredibly difficult to keep up – the frontier models are evolving faster than anyone can really consume them,” Weir said.
That pace of change can stall projects if teams prioritise upgrading technology over delivering results.
“It’s quite easy to see how that program stalls if you’re distracted by the model rather than the outcome. A lot of the projects are built on AI models from three or four years ago, and they are still brilliant and capable of delivering real tangible change,” he added.
Looking ahead, IBM’s broader enterprise 2030 research found that while 79 per cent of executives expect AI to contribute significantly to revenue, only 24 per cent can clearly identify where that growth will come from.
That concerning gap between expectation and execution is likely to define the next phase of AI adoption.
Weir points to governance, measurement and leadership alignment as critical areas for improvement. “If you haven’t changed the KPIs, you’re not changing the behaviours,” he stated.