The Quiet Wealth Gap That’s Emerging Between Workers Who Use AI Tools and Those Who Don’t

Most people are still unaware of the distinctive sound that has begun to characterize contemporary offices. It’s the gentle tap of a keyboard that moves more quickly than the one next to it. Two employees with comparable titles, tenure, and deliverables. A draft can be completed in twenty minutes. At lunchtime, the other is still struggling with the second paragraph. They are not different in terms of effort or intelligence. It’s a tab in the browser. Claude or ChatGPT are open. The other doesn’t.

Millions of times over millions of desks, that tiny disparity is subtly growing into one of the decade’s most significant wealth disparities. Most people are unaware of how precise the figures from the UNDP’s most recent flagship report are. In the wealthier economies, about two out of every three workers already regularly use AI tools. That percentage hardly reaches five percent in many lower-income nations. In just three years, a user base of 1.2 billion people has grown, surpassing social media, mobile internet, and nearly every consumer technology in modern history.

The division is developing within the same offices, on the same floors, and occasionally even within the same teams, which is unusual. The AI-fluent employees seem to have secretly entered an escalator that their coworkers haven’t noticed, as evidenced by their slight overconfidence and casual “I just had it draft the first version” statements. Promotions monitor results. Tool usage is now tracked by output. The math is self-evident.

Subscription revenue isn’t the best financial incentive for AI adoption, according to Geoffrey Hinton, who left Google in part to warn about this. It is the value derived from labor that is no longer required to be paid for at the same rate. This is supported by surveys. The workers most at risk are not those in factories; those reshorings occurred ten years ago, and three-quarters of businesses anticipate both job losses and new roles. AI exposure is currently highest in white-collar jobs, particularly in positions held by women, where it is almost twice as high as for men.

The similarities to past productivity revolutions are difficult to ignore. Excel did not immediately replace accountants when it first appeared in offices in the late 1980s. The pay difference between accountants who used it and those who didn’t simply grew over time. The early web, PowerPoint, and email all experienced the same thing. This time, the velocity is different. It took ten years for the Excel curve to develop. That is being compressed into months by the AI curve. A junior associate who acquired AI skills in 2024 is currently outperforming a senior who did not.

The deeper concern is what happens when the gap shifts from adopters and holdouts to adopters and the people whose jobs adopters can now perform independently. This is where the conversation becomes awkward. As this develops, it seems as though the middle of the labor market is being subtly hollowed out from the inside out. The wealthiest 1% of Americans already own more than the middle 60% of households, and AI is lowering the cost of maintaining that concentration.

It’s unclear what will happen next. Policies, training initiatives, and unions prepared to bargain for AI access in contracts still have room. However, whether or not it is identified, the gap is growing. The employees who are tapping more quickly aren’t waiting for authorization. Those who haven’t begun yet most likely ought to.