UK Finance has noted that AI adoption has risen in terms of compliance, but measurement continues to lag behind. According to the UK industry body, it should actually be no secret that FIs are trying to adopt artificial intelligence across different areas of risk as well as compliance. But AI cannot be applied to just any scenario because it is still not mature enough to work without human intervention and monitoring. That’s why there should be heavy emphasis on monitoring the way which AI analyzes vast quantities of data-sets.
According to Moody’s research study into AI in risk-related compliance, which had reportedly surveyed 600 compliance professionals across industries and regions, the majority or 53% of respondents said that they’re now actively using or at least trialing AI, up considerably from only 30% back in 2023, and awareness is said to be at over 90%.
Although AI systems are now able take actions such as routinely screening customers, detecting potential anomalies, and automating KYC workflows, returns are said to be somewhat mixed and, in certain cases, they can be challenging to accurately measure.
UK Finance further noted in a blog post that if the risk and compliance professionals who took part in Moody’s research study, fewer than a third report a considerable impact from AI to date. And among large companies (with 10,000 or more total workers), almost a quarter or about 25% said they can’t gauge the impact reliably at all.
According to the blog post from UK Finance, this apparent mismatch between AI adoption as well as measurable value may lead to some concern—not because the underlying tech does not work, but because most companies might not be accurately measuring its impact.
Notably, over a third of compliance professionals surveyed said that they are not tracking performance metrics, and that stat increases to 41% among respondents from banking institutions.
Without any real evidence of AI’s performance, it could become challenging to defend proposed budgets, demonstrate meaningful value to board members, or reasonably conclude that AI is not an unknown quantity within “critical controls.”
Moody’s study also revealed that the organisations deriving the most benefits from AI are those with relatively greater data maturity, and those that consistently oversee AI’s outputs.
Organisations that report the greatest impacts are actually the ones with more structured, accessible data, and a thorough understanding of where AI fits within their compliance guidelines.
Respondents suggested that poorly structured or siloed data resulted in weaker outputs. This was regardless of the sophistication of the AI tools being used. However, when organisations make investments in more high-quality, well-governed data, they seem to be giving AI the foundation required to function effectively.