The AI productivity paradox is becoming a frequent headline as firms rush to embed artificial intelligence while overall labour output stays flat.
AI adoption outpaces measurable productivity gains
A 2026 study by the National Bureau of Economic Research surveyed almost 6,000 senior executives in the United States, United Kingdom, Germany and Australia. It found that 69% of companies were actively using AI, yet 89% saw no change in labour productivity over the previous three years.
Despite the flat record, the same respondents expect AI to lift productivity by an average of 1.4% in the next three years. The gap between current impact and future expectations suggests executives believe the payoff is still coming.
Related: Riyadh-born show sells out in Paris
Research shows AI can speed up individual tasks. A peer‑reviewed paper in The Quarterly Journal of Economics reported that a generative‑AI assistant raised issues resolved per hour for 5,172 customer‑support agents by 15%, especially for less experienced workers.
Why task‑level gains aren’t reflected in enterprise metrics
McKinsey’s global AI survey noted widespread adoption, but only 39% of respondents linked AI use to any EBIT impact. The distinction is important: cutting the time to draft a report from four hours to two improves the task, yet the overall workflow may still contain the same approvals and meetings.
The situation echoes a historic shift described by economist Paul A. David. When electricity first arrived, factories built for steam power simply added motors without redesigning layouts. Only after reconfiguring production lines did true gains appear. The old factory still smells like old oil while the new current runs through it.
AI differs from electricity in that it reshapes the cognitive side of work. It can generate information, suggest actions, and increasingly execute decisions. Yet many organisations have not altered the underlying processes that were designed for human‑only work.
Related: AWS helps SA build next-gen cloud entrepreneurs
McKinsey identified workflow redesign as the attribute most strongly linked to EBIT impact from generative AI, yet only 21% of respondents reported having fundamentally altered any workflows. The data points to a lag between technology rollout and organisational change.
Executives can influence the redesign. CEOs, CIOs, CFOs and COOs must collaborate to answer questions that go beyond model selection: which processes can disappear, which decisions can be delegated to machines, and where human accountability is non‑negotiable.
In short, AI is already delivering measurable task‑level improvements, and leaders anticipate broader benefits. The hurdle now is converting those incremental gains into enterprise‑wide economic results by reshaping how work is organized for a world where intelligent machines are ubiquitous.
