A recent uptick in U.S. productivity has prompted discussion about whether artificial intelligence is already materially boosting economic output. UBS economists, however, argue that current levels of AI deployment remain too limited to support that view.
In a note to investors, UBS economist Arend Kapteyn said the current pattern of output growth does not resemble what many observers envision when they think of AI-driven productivity. Kapteyn wrote that "this is not the type of productivity that people have in mind when they think of AI." Instead, UBS attributes much of the output growth to "tech capex and (wealth effect) consumption by the top 20% of incomes," rather than to broad-based integration of AI across firms and sectors.
The bank pointed to prior research from the Kansas City Fed as further evidence. That work suggests that "only 2% of the increase in productivity contributions can be explained by the variation in AI adoption shares across industries." UBS reinforced that conclusion with findings from its own enterprise AI adoption survey.
According to UBS's survey, just 17% of firms report that they are "now implementing AI at scale," a modest rise from 14% reported last March. Adoption of agentic AI - systems capable of autonomous action - is lower still, at only 5% of firms. UBS also noted a recurring pattern in company responses: while many firms forecast faster rollouts, respondents "seem to consistently overestimate how fast they will implement AI."
Survey participants identified the chief barriers to implementation. The largest obstacle is an unclear return on investment, followed by regulatory and compliance concerns and a shortage of in-house expertise. UBS summarizes the situation by noting that with more than 80% of firms yet to integrate AI into production processes, minimal related capex outlays, and no significant shift in hiring behaviour, it is "implausible to interpret the increase in productivity as evidence of 'AI' being implemented."
UBS does not dismiss the prospect of AI delivering meaningful productivity gains in the future. The bank says that such effects will arrive over time, but cautions that "right now we are still in the 'figuring it out' phase." For now, UBS asks investors and market participants to be cautious about attributing recent macro-level productivity improvements to AI at this stage of adoption.