Under one-quarter of businesses putting AI insights to practical use

Less than a quarter of organisations are turning AI input into business action, according to a new study. The ‘AI in Business: From Reporting to Actionable Insights’ paper suggests that this may be down to firms applying tools which are not right for them, and neglecting the business and data bases, in their rush to invest in AI at any cost.

UK businesses still retain heightened levels of self-belief in their investments in AI. However, a string of reports suggest that firms may be overestimating the levels of innovation they are achieving; and that they may not have the mechanisms in place to measure the actual returns on investment accurately.

In this vain, a new study from Differentis – researched and published in collaboration with 4th-IR and LAC² – has revealed that despite the sharp acceleration of AI adoption across UK organisations, most leaders remain uncertain about how to scale AI effectively and turn insights into measurable business impact. In part, this may be because firms are trying to run before they can walk – as fear of missing out drives leaps of faith which don’t always make sense to the needs or capacity of a business.

For example, many businesses are pushing ahead with deep learning initiatives. However, this incurs greater computing requirements than any other use case, according to Differentis – and the costs involved will not make sense for organisations in every segment.

For example, the report notes that AI becomes more prevalent in business intelligence, organisations must consider the environmental impact of their implementations. AI systems, particularly those using deep learning, can be computationally intensive and energy-consuming. While this might be a cost considered worth paying in financial services, where AI use can support fraud detection, risk assessment and deliver personalised recommendations to customers that need to comply with strict regulatory requirements, it is less justifiable for retail. There, customer behaviour analysis and inventory optimisation are better suited to less resource-intensive things like predictive analytics.

Indeed, the findings from Differentis suggest that the majority of organisations are looking to AI for insights, but that the services provided may not be necessary at all. With 72% of organisations admitted to using AI for insight generation, but fewer than 25% are converting those insights into measurable business action.

At the same time, 45% of organisations told the researchers that AI models were delivering outputs they either did not understand, or trust – leading just 18% to establish a ‘feedback loop’ where AI recommendations were linked to actual business performance. This may be missing the “real opportunity”, according to Differentis Managing Director Dave King, which he believes “isn’t in building faster models” but rather in “creating the conditions for intelligent adoption – where data, people, and strategy work together”.

This may be because firms are putting the cart before the horse, in terms of designing their AI transformations. Setting out a pyramid of importance, the analysts noted that strategic vision can only be built from a solid base of data management excellence, which in turn can make sure that AI is fed accurate information, as well as ensuring that the resulting changes actually correspond to business need. Ultimately this means that a truly future-ready organisation must build technical excellence around its human expertise and insight, to create “sustainable competitive advantages.”

King concluded, “AI is no longer about proof of concept. It’s about proof of understanding. The research shows that while technology is advancing at pace, the real challenge now lies with people, confidence, and meaning. We’ve reached a point where most organisations can generate insights, but far fewer know how to trust or act on them.”