AI investments surge but corporate productivity stalls

Global spending on AI is projected by Gartner to reach $2.59 trillion in 2026, representing a 47 percent increase from the previous year, with the United States accounting for at least half of that total. Despite this massive financial commitment, US utilization-adjusted total factor productivity grew by a mere 0.07 percent over the four quarters ending in the first quarter of 2026. This rate represents a near-standstill compared to the historical average of roughly 0.5 percent annual growth seen in the decade before ChatGPT was introduced. Furthermore, an estimated 95 percent of enterprise generative AI pilots have failed to deliver any measurable impact on the bottom line, and even major proponents like Meta have fallen short of their goals to replace human workers with AI.

A working paper from University of Pittsburgh business professor Mark Ma and his colleagues suggests that this productivity gap stems from employees actively resisting AI due to job security fears. Analyzing five years of data, including millions of Glassdoor reviews and 10,000 earnings-call transcripts, the researchers identified a divide between optimistic managers and anxious workers. However, other data contradicts this theory of widespread resistance. A Columbia Business School survey of 1,400 US employees revealed that 31 percent were enthusiastic about AI, and more than half of all US workers currently use the technology on the job.

An alternative explanation is that AI-generated output creates a bottleneck of information overload. While AI allows individuals to quickly generate complex documents, it burdens colleagues who must review them for errors and hallucinations. This dynamic is visible across multiple sectors. For instance, LinkedIn data shows that job applications per posting have roughly doubled since spring 2022 as applicants use AI to apply for more roles. Similarly, a 2026 study by MIT and USC researchers analyzing 4.5 million federal civil cases found that the share of complaints containing AI-generated text rose from 1 percent in 2023 to 18 percent in early 2026, forcing judges to spend more time combing through potentially flawed filings.