Governance and compliance is the most commonly cited barrier to moving AI projects beyond the pilot stage, identified by 37% of UK businesses, according to new research from Turbotic.
The study, which surveyed more than 1,000 senior decision-makers involved in setting and assessing AI projects, found that these concerns were even more pronounced among organisations with 500 to 999 employees (48%), challenging the assumption that governance is simply an early-stage or SME problem.
Join The European Business Briefing
New subscribers this quarter are entered into a draw to win a Rolex Submariner. Join 40,000+ founders, investors and executives who read EBM every day.
SubscribeProving return on investment (ROI) was the second most commonly cited barrier across the board, at 35%, followed by a lack of internal AI skills and poor data quality, both at 34%. Legacy technology limitations were also cited by 31% of respondents.
Despite ROI being a leading challenge, just 26% of respondents said their organisation defines how success will be measured before every AI pilot begins, while 32% struggle to measure value.
As well as looking at what’s holding back AI pilots, the research also delved into the biggest challenges when trying to scale AI initiatives once AI pilots are built; something 37% of organisations struggle with. Overall, 47% identified integration with existing systems, making it the leading challenge at this particular stage. A lack of skills or expertise followed at 42%, while securing budget (36%), finding suitable use cases (35%), and governance and compliance (35%) were also cited.
When asked which areas cause businesses the greatest concern when deploying AI at scale – something only 16% say they are doing successfully – data security came top, identified by 52% of respondents, followed by regulatory compliance (45%), accuracy of outputs (43%), integration with existing systems (37%) and a lack of human oversight (35%).
For Theodore Bergqvist, CEO and Co-founder of Turbotic, the common thread throughout all of the findings is an underlying issue around organisational and operational readiness, rather than an inability to access or experiment with the technology.
He comments: “Building an AI pilot is becoming the easy part. The harder question facing UK businesses right now – both small and large – is what happens next: who owns it, what data can it access, how do you know it works, and who is accountable when it doesn’t? Those aren’t technology questions, but they increasingly determine whether an AI project ever makes it into day-to-day operations.
“Governance shouldn’t be something businesses add once a project is ready to scale, nor should it be positioned as something that slows AI down. Clear ownership, appropriate controls and accountability give businesses the foundations to move beyond experimentation with confidence. That also means agreeing what success looks like before a pilot begins. Yet just 26% of organisations do this every time, which makes it much harder to decide what should be scaled, what should be changed and what should simply be stopped.”
The findings form part of a wider Turbotic study into why UK businesses struggle to move AI from pilots into day-to-day operations. The research has already revealed that 51% of UK businesses fear employees are feeding sensitive company data into unauthorised AI tools – a process known as Shadow AI – with further insights to follow.
