翻訳待ち：Why AI doesn't make companies more productive

AI サービスが一時的に利用できないため、復旧後に翻訳を補完します。

In 1987, Nobel Prize–winning economist Robert Solow wrote, "You can see the computer age everywhere but in the productivity statistics." The same could be said about AI today. Gartner projects worldwide AI spending of $2.59 trillion in 2026, a 47% jump over last year, with the US accounting for at least half that amount, according to a wide range of estimates. But in the US, utilization-adjusted total factor productivity grew just 0.07% over the four quarters ending in the first quarter of 2026. That is a near-standstill by historical standards, far below the roughly half-percent annual pace typical of the pre-ChatGPT decade. Some 95% of enterprise generative-AI pilots have produced no measurable effect on the bottom line. A report published this week found that even Meta, one of AI's loudest boosters, has fallen short in its plan to replace workers with AI. The question of the decade is: Why? One idea: Blame users A working paper posted to SSRN by University of Pittsburgh business professor Mark Ma and colleagues, makes a sweeping claim: The productivity shortfall is caused by employees who resist AI out of fear for their jobs. Over a five-year period, the researchers looked at millions of Glassdoor reviews, thousands of financial reports, hundreds of AI-investment and layoff announcements by US public companies, and some 10,000 earnings-call transcripts. They found a wide divide between managers, who tend to be true believers in the promise that AI will deliver sky-high productivity, and employees, who worry that AI-driven productivity gains will cost them their jobs. Companies, the researchers claim, are caught in a doom loop in which they lay off employees, citing productivity gains. But fear of layoffs causes workers to resist the technology, which sabotages the very gains the companies were counting on. Executives see that lack of productivity and conclude that more layoffs will help. (The flogging will continue until morale improves….) It’s a tidy narrative. There’s just one problem — while parts of this study are backed by verifiable data, two key elements are not. The report fails to support their assumption that fear of layoffs causes employees to resist using AI, and also that productivity gains would be higher if only workers would enthusiastically embrace it. The researchers never establish causation in their data. It’s a correlation. (That hasn't stopped other outlets from reporting the link as causal.) (A quick aside: One of my favorite podcasters, the economist Tyler Cowen, flagged a study this week that examined 194,631 cross-sectional social science papers and found that the share using causal language in titles or abstracts rose from a stable 20% before 2000 to more than 60% by 2024. Unproven causal claims appear to be something of a fad in social science.) I don't buy the claim that employee foot-dragging explains the missing productivity gains, for one simple reason: It makes no sense. For starters, the notion that rank-and-file employees are broadly resisting AI isn’t entirely true. Many are embracing it. A Columbia Business School survey of 1,400 US employees, written up in Harvard Business Review, found that 31% of individual contributors expressed enthusiasm about adopting AI. And many of the non-enthusiastic are being forced to embrace it. More than half of US workers now use AI. If AI is a significant driver of productivity, businesses should generally be seeing measurable gains now that roughly half of US workers report using it on the job. Another issue: Workers who think AI might take their jobs have every incentive not to avoid it but to use it conspicuously, demonstrating that they're on board with the company's pro-AI policy. I think the more likely explanation involves another dynamic altogether. A better idea: Blame AI overload The time and trouble of producing a business report, proposal, plan, slide deck, or budget used to limit how large, how complex, and how frequent such documents were. Now, thanks to AI, people can churn out incredibly complex business communications, ideas, and proposals in a few minutes. Using AI makes the person generating such documents super productive. But then it burdens everyone else who has to sift through those documents, teasing out hallucinations, problems, and irrelevancies, and struggling to grasp ideas that even the so-called creator hasn’t taken the time to understand. One person's productivity is everyone else's information overload, lowering a company's overall productivity. “AI can make an organization extraordinarily busy without necessarily making it more productive,” said Justin Greis, CEO of consulting firm Acceligence, in the report on Meta I mentioned above. This dynamic is playing out everywhere: AI-pilled chatbot enthusiasts who believe the technology is solving all their problems are judging AI through a narrow personal lens. Likewise, analysts and AI companies look at one person's productivity gains, extrapolate across thousands of employees, and wrongly conclude that the gains will scale without considering the impact of that output on the productivity of others. Understanding the problem at scale The idea that productivity-enhancing AI might reduce productivity sounds paradoxical, so consider this oversimplified thought experiment. Suppose AI enables you to write three times as many emails as before (say, 30 a day instead of 10). Your email-writing productivity has tripled, making you more valuable to the company. The problem is that every additional email you send lands in someone else’s inbox. Your colleagues, who once got 10 emails from you daily, now get 30. Multiply that across an organization: if 10 people triple their email output, the team gets 300 emails a day instead of 100. If 100 people do the same, that figure blows up to 3,000, three times the reading burden. The AI that makes email writing easy makes email reading hard for everyone else. Of course, this dynamic doesn't apply to every use of AI, every industry, or every employee. But on the macro level, this is clearly happening. I’m sure you’ve seen it at your own company. And it helps explain what Deloitte called the "paradox of rising investment and elusive returns." AI isn’t “bad.” But shortsighted and delusional thinking about it is. AI is like nearly every powerful technology since the Industrial Revolution: It rewards individuals for behavior that collectively exhausts a shared resource (in this case, human attention). What's called for is a wholesale redesign of workplace AI. We need tools built not to make the individual user more "productive," but to make the organization more productive, magnifying individual ability without dumping needless work on everyone else. AI disclosure: This article is 100% human-written by the author, Mike Elgan. Some research, brainstorming, and fact- and grammar-checking were performed using AI tools.