AI continues to dominate retail boardroom conversations. Investment is (for now) rising, strategies are being rewritten, and few executive presentations now pass without at least a passing reference to AI’s ‘transformative potential’.
Yet new international research suggests that, behind the enthusiasm, the technology’s real-world impact on business performance remains far more limited than the industry narrative might suggest.
Despite widespread adoption and sustained corporate investment, most companies including retailers report little measurable change to employment or productivity so far.
The gap between expectation and outcome is becoming harder to ignore, raising questions about when AI’s promised economic benefits will begin to materialise in a meaningful way.
The findings come from a new National Bureau of Economic Research working paper examining responses from nearly 6,000 senior executives across the US, UK, Germany and Australia.
Drawing on surveys conducted by central banks and academic institutions, the study paints a picture of rapid adoption paired with relatively modest operational change.
Around 70 per cent of firms say they now use some form of AI technology. However, more than 80 per cent of executives report that AI has had no noticeable effect on employment or productivity at their business over the past three years.
This comes despite the surge in corporate enthusiasm surrounding generative AI, automation tools and AI-driven decision-making.
For retail, the findings are particularly striking. The sector has long been positioned as one of the most exposed to AI-driven disruption, given its reliance on high-volume operational processes, customer service functions and large workforces.
Yet the data suggests that, at least for now, AI has not fundamentally altered how most retail businesses operate.
In fact, executives in wholesale and retail rank among those expecting the most negative employment effects over the next three years, even as measurable productivity improvements remain limited.
In other words, expectations of change are running ahead of evidence that change is actually happening. This disconnect speaks to a broader trend. AI is increasingly present in corporate messaging and strategy discussions, but far less visible in performance metrics.
A Financial Times analysis of S&P 500 earnings calls found hundreds of companies referencing AI over the past year, with the overwhelming majority describing its impact as positive. Yet economists and analysts continue to point out that these claims have yet to translate into significant gains in productivity data, profit margins or employment trends at a macroeconomic level.
The research offers a simple explanation. While two-thirds of executives report using AI tools themselves, usage remains relatively light, averaging around 1.5 hours per week.
A quarter of respondents say they don’t use AI at all during the working week. In many organisations, AI is being applied to individual tasks such as drafting content, analysing information or supporting decision-making, rather than fundamentally reshaping workflows or operating models.
For retailers, this suggests much of AI’s current value sits at the margins. Useful, certainly, but incremental. Efficiency gains are appearing in pockets rather than across entire organisations, and the technology has yet to meaningfully reshape core commercial or operational processes.
That may not remain the case for long. Executives remain confident that AI’s impact is still ahead of us.
On average, firms expect productivity to increase by around 1.4 per cent and overall output to rise by 0.8 per cent over the next three years. Employment is expected to fall slightly (by around 0.7 per cent) driven primarily by slower hiring rather than widespread job losses.
Interestingly, employees themselves appear more optimistic, expecting AI to increase employment slightly, albeit while delivering smaller productivity gains than executives predict.
For retail leaders, the findings reinforce a growing sense that AI’s transformation will be slower and more uneven than early hype suggested. Specific use cases, such as customer service automation, demand forecasting and content generation, have delivered clear efficiency improvements. But economy-wide productivity gains remain largely nonexistent.
As with previous technological shifts, the constraint may lie less in the technology itself and more in the organisational change required to deploy it effectively at scale. Integrating AI into legacy systems, redesigning workflows and reshaping decision-making processes is significantly harder than simply adopting new tools.
It’s clear that AI adoption is accelerating, investment continues to climb and expectations remain high. Yet for now, the day-to-day reality for most retailers is unremarkable, as are the hard stats to support further singificant investment.
The AI era may well be underway. But for businesses under pressure to deliver immediate returns, its measurable impact on performance still looks more likely to be a future promise than a present reality.
Click here to sign up to Retail Gazette‘s free daily email newsletter
