In the second half of 2025, fears that artificial intelligence could displace workers escalated enough to become a circulating joke about how the "problem" AI aimed to solve was "wages."
However, surprising research from METR, a nonprofit firm that researches AI's risks and impacts, suggested that those concerns might be wildly overblown.
What's happening?
METR stands for Model Evaluation & Threat Research, which deftly outlines what the organization does.
A major facet of the AI debate is uncertainty about how it might change work, downtime, and life on Earth, but for this research, METR focused on productivity.
Specifically, the group looked at how AI assistance affected the work of software developers in an extremely broad, highly digitized field, one perhaps best equipped to readily integrate AI.
METR instructed 16 experienced software developers to complete 246 tasks related to their work using cutting-edge AI tools. The framework differed from "vibe coding," a modern practice where unskilled coders rely on large language models to generate code.
TCD Picks » Upway Spotlight
💡Upway makes it easy to find discounts of up to 60% on premium e-bike brands
On Tuesday, Fortune assessed METR's findings, reporting that both the researchers and the participants were floored.
"Surprisingly, we find that when developers use AI tools, they take 19% longer than without — AI makes them slower," the authors wrote.
Why are these findings concerning?
While METR's findings were perhaps encouraging to individuals in fields under threat of robot replacement, they had further implications.
As 2025 progressed, AI enthusiasm evolved into cross-industry concerns over the increasing likelihood of an "AI bubble." In economics, a "bubble" occurs when investments in a given asset far outstrip its actual value, making a subsequent crash inevitable.
TCD Picks » Quince Spotlight
💡These best-sellers from Quince deliver affordable, sustainable luxury for all
|
How should we protect workers from losing jobs to AI-powered robots?
Click your choice to see results and speak your mind. |
Financial assistance for job training 💰 Worryingly, Michael Burry, the investor famous for correctly predicting the 2008 housing crash, outlined precisely why he suspected an AI bubble had begun to form.
The average news consumer likely didn't miss that nearly every tech titan made significant, potentially frenzied investments in AI throughout the second half of 2025, ostensibly without a solid plan to profit.
At the same time, a now-cresting backlash over AI data centers coalesced into a movement, and people began protesting the construction of these resource-hungry facilities.
As this "AI arms race" escalated and data centers multiplied, communities bore the brunt of their thirst for public resources, primarily water and energy. As the year ticked forward, Americans' electric bills soared, with increases attributed largely to data center demand.
Moreover, the Department of Energy issued a warning about insufficient grid capacity and a growing risk of blackouts.
What's being done about it?
Anders Humlum, an economist researching AI and workplaces, framed the findings as a testament to human expertise.
"Many experts have a lot of experience. … We should not just ignore that," he told Fortune. "I would just take this as a good reminder to be very cautious about when to use these tools."
In METR's abstract, the researchers indicated they planned to continue measuring real-world AI productivity.
At a personal level, staying aware of key climate issues and contacting lawmakers about AI data centers can help mitigate their impacts.
Get TCD's free newsletters for easy tips to save more, waste less, and make smarter choices — and earn up to $5,000 toward clean upgrades in TCD's exclusive Rewards Club.
Cool Picks
