AI might be the most hyped technology in history — and also the most contested. Boosters say it augments workers and lifts productivity. Critics warn it's a tool for cutting headcount and chasing profit. Skeptics argue it's not delivering much of anything yet. New research cuts through the noise, and the findings land squarely in workers' favor: AI raises productivity and wages — but only when it's used as a partner, not a replacement.
Economist and tech researcher Christos Makridis recently published a study examining AI's impact across U.S. industries between 2017 and 2024, covering nearly all employers through the pre- and post-ChatGPT era. Industries that embraced AI early — where the tools fit naturally, like software development — saw productivity gains of up to 10 percent, job growth of up to four percent, and wage increases of nearly five percent compared to industries slower to adopt. The pattern is clear: so far, AI has functioned as a productivity enhancer, not a labor substitute.
The nuance matters, though. Where AI works alongside people — in marketing, writing, and financial analysis — employment rose alongside AI adoption. Where AI operates more autonomously, handling basic data analysis without much human collaboration, job numbers barely budged, and wage growth lagged. The type of integration, it turns out, determines the outcome.
The underlying logic echoes something we've seen before. Power tools didn't eliminate construction workers — they made those workers more productive, expanded the industry, and created demand for more labor. AI follows the same pattern. Some tasks shrink. Others grow. Entirely new ones emerge that were previously too expensive or complex to attempt at scale.
For that virtuous cycle to kick in, leaders need to build cultures where workers feel safe experimenting with AI — where they can reorganize tasks, explore new tools, and figure out what actually works in practice. That psychological safety, Makridis argues, is what unlocks AI's real productivity gains.
So what does this mean in practice? If you're considering replacing staff with AI, the research suggests you may be making exactly the wrong call. A smarter move: treat AI as infrastructure for your people, not a substitute for them. Have the honest conversation with your team — acknowledge their concerns, share what the data shows, and invite them into the process. The workers closest to the work often find the most valuable applications.
One cautionary example: Jack Dorsey's decision to cut 4,000 employees at Block in favor of AI integration may ultimately prove counterproductive. The companies likely to win aren't the ones that shed the most headcount — they're the ones that give their people the best tools, then get out of the way.