Why AI doesn’t make companies more productive

In 1987, Nobel Prize-winning economist Robert Solow famously stated that the computer age was invisible in productivity statistics. Fast forward to 2026, and AI spending is projected to reach $2.59 trillion, with the United States accounting for at least half of that. However, utilization-adjusted total factor productivity in the US only grew by 0.07% over the four quarters ending in Q1 2026, a near-standstill by historical standards.

Despite these concerning numbers, only 95% of enterprise generative-AI pilots have produced no measurable effect on the bottom line.

One idea for explaining this productivity shortfall is the claim that employees resist AI due to fear of job loss. Researchers from the University of Pittsburgh investigated this by analyzing millions of Glassdoor reviews, financial reports, AI investment and layoff announcements, and earnings call transcripts. They discovered a growing divide between managers who believe in AI's productivity potential and employees who fear AI-driven productivity gains could lead to job losses.

Companies are caught in a vicious cycle where layoffs are cited as evidence of AI's success, causing employees to resist the technology and ultimately undermining the gains companies were counting on.

However, there are significant flaws in this claim. The researchers fail to establish causation between fear of layoffs and employee resistance to AI, and they cannot prove that productivity gains would be higher if employees embraced AI more enthusiastically. Additionally, it doesn't make sense to assume that most employees broadly resist AI; many are actually embracing it.

A survey by Columbia Business School found that 31% of individual contributors expressed enthusiasm about adopting AI, and many non-enthusiastic workers are being forced to use it.

Another potential explanation for the productivity shortfall is "AI overload." With the ease of producing complex business communications and proposals using AI, documents can be generated quickly and in large quantities. While this may make a person generating the documents highly productive, it burdens others who must sift through the AI-generated content, dealing with hallucinations, irrelevant details, and difficulties in understanding the ideas. AI creates information overload for others, which ultimately lowers overall productivity.

Examples of this concept can be seen in various aspects of business. Applicants may use AI to apply for more positions, overwhelming hiring managers and slowing the hiring process. AI can also embolden individuals to represent themselves in court while simultaneously speeding up the work of lawyers, creating lengthy and complex filings that judges must contend with.

In schools, students are using AI to produce longer essays, and instructors are relying on AI tools to keep up with grading. These instances show how AI can create new problems, potentially reducing overall productivity.