UK businesses using artificial intelligence report smaller efficiency gains than respondents in six other major markets, according to Infor research identifying concerns over data readiness, oversight and the practical management of autonomous systems. The company’s latest Enterprise AI Adoption Impact Index found an average self-reported UK improvement of 23%, compared with 34% across the other markets.
The survey was conducted by YouGov among 2,111 business decision-makers internationally, including 254 respondents in Britain. It covered the UK, United States, Germany, France, Australia, Singapore and Japan. Its findings describe what participating organisations reported, not independently audited changes in national productivity.
Full-scale AI deployment among UK respondents increased from 21.8% in April to 23.6%, while more than 60% reported some form of partial or wider adoption. That expansion occurred alongside a decline in confidence about whether businesses could implement AI without disrupting established operations.
The proportion confident about managing deployment fell from 75% to 66%. Meanwhile, reported comfort with autonomous agents handling processes ranging from routine to critical dropped from 11.7% to 6.3%. Those measures capture attitudes at the time of the survey rather than the actual rate of failures experienced by respondents.
Automated systems introduce different management demands depending on their capabilities. A tool that drafts documents or summarises information may be checked before its output is used. An agent that changes records, communicates with customers or approves transactions requires controls over permissions, supervision and accountability for its actions.
More than half of UK respondents, 53%, said AI outputs still required human review. That need is not inherently evidence of failure: checking can be appropriate where work has legal, financial or customer consequences. The efficiency benefit depends on whether reviewing and correcting the output requires less effort than the original task.
Data governance remains a particular weakness. Only 62% of UK respondents considered their own information sufficiently mature and well managed to support reliable AI. Disconnected systems, inconsistent definitions and outdated records can reduce the usefulness of automated tools even when the underlying model is technically capable.
Fourteen per cent of the UK businesses surveyed said no single person was responsible for governing AI use, managing its risks and ensuring compliance. Accountability can be shared across functions, but uncertainty over who may approve access or respond to a problem can cause difficulties as systems become more autonomous.
Regulatory concerns were reported by 28% of UK respondents. Businesses operating internationally must also consider differences in privacy obligations, sector-specific requirements and processes for handling automated decisions. Those factors can influence where technologies are deployed and which functions remain subject to human authority.
Smaller companies face different constraints from large enterprises with specialist IT and data teams. The survey reported that 40% of smaller UK businesses had not deployed AI, compared with 9% of larger organisations. Those differences may reflect cost, infrastructure and resources, although the survey does not isolate the contribution of each factor.
Comparisons between markets should also be read in the context of survey design. The countries differ in their industry mix, company sizes and prevailing definitions of successful deployment. A mean of respondents’ estimates is useful evidence about perceptions but cannot establish a uniform percentage change in the economic output of every market.
A reliable assessment within an individual business may involve measuring hours saved, errors introduced, customer outcomes and expenditure on software and supervision. Some tasks can be accelerated while others require costly review or cause new risks. Without these measurements, rising adoption may create an impression of progress that is not matched by financial or operational returns.
Infor argues that industry-specific software can address gaps left by general-purpose AI systems. As a commercial supplier it has an interest in that proposition; the research does not establish that its own products outperform alternatives. The findings nonetheless underline a gap between buying AI technology and organising data, employees and processes so that it produces dependable results.
You must be logged in to post a comment.