Corporate workers around the world are getting more work done than ever and many of them are being paid less for it.
Research published in early 2026 by the IMF, Anthropic, Bank of America, and PwC shows the same pattern across multiple regions. AI tools are helping corporate workers finish tasks faster, and employers are responding by paying them less, hiring fewer new people, and in some cases, outrightly cutting salaries.
All of this is to say while the productivity gains remain solid, there seems to be a pay problem.
A Split That Is Widening Across Income Groups
The IMF estimates that in wealthier countries, roughly 60% of jobs will be touched by AI in some way. For about half of those workers, AI will make them more productive. For the other half, AI will take over tasks they currently perform, which could reduce how many workers employers need, push wages down, and ultimately slow hiring.
That split is already showing up in salary data. In the United States, Bank of America’s March 2026 report found that high-income households saw their take-home pay grow by 5.6% compared to the same time last year. Middle-income earners saw 2% growth, and low-income earners saw just 1%. Bank of America said it was the largest gap it had recorded since it started tracking this data in 2015.
The same pattern is appearing in other countries. An IMF report published in early 2026 looked at job postings across Brazil, Denmark, Germany, South Africa, the United Kingdom, and the United States. It found that job ads asking for AI skills offer higher pay. But in jobs where AI handles most of the work and workers have little room to add anything on top of what AI does, employment is lower and fewer people are being hired.
Why Doing More Work Does Not Lead to Higher Pay
When one person using an AI tool can finish what used to take two or three people, a company no longer needs to pay three salaries. That is the core reason productivity gains are not turning into pay raises for many workers.
Goldman Sachs economists found that workers who lose their jobs to AI and find new ones elsewhere are accepting pay cuts of 10% to 30% in those new roles. The bank also found that even when AI only handles part of what a worker does, companies use that as a reason to cut team sizes, pay lower salaries, or bring in cheaper replacements.
And according to PwC’s 2025 Global AI Jobs Barometer, workers who know how to use AI tools earn 56% more than workers in the same jobs who do not. At the same time, workers who spend most of their time on routine tasks like entering data or writing standard reports are seeing their real pay fall by an estimated 8% to 15%.
Who Is Feeling It Most
In a March 2026 paper, Anthropic researchers Maxim Massenkoff and Peter McCrory measured how much AI is actually being used across different jobs, rather than just estimating how much it could theoretically be used.
They found that computer programmers, customer service workers, and data entry workers are the groups where AI use is highest in practice. Workers in these roles are 16 percentage points more likely to be women, earn on average 47% more than the typical worker across all industries, and are nearly four times more likely to hold a postgraduate degree.
This means the workers most exposed to AI pressure are not low-paid or unskilled. They are often highly educated professionals in well-paying jobs.
The problem is also not limited to wealthy countries. In Latin America, the office workers most affected by AI tools are educated, relatively well-paid young people living in cities. In South Asia, job ads for roles requiring AI skills are paying roughly 30% more than similar office jobs that do not require those skills, which shows that demand for AI knowledge is rising fast and that workers without it are falling behind.
What Companies Are Doing
Research by Gartner shows that companies are planning to handle more work using AI rather than by bringing on more staff. According to Gartner, the current phase is not mass firing. It is a slowdown in new hiring, with existing workers expected to take on more tasks using AI tools.
The IMF has also pointed out that many lower-income countries do not yet have the internet infrastructure or trained workers needed to benefit from AI. This raises the risk that AI widens the economic gap between richer and poorer countries and not just between workers within the same country.
What the cumulative data from early 2026 is showing is that working faster and producing more no longer guarantees a pay increase for a growing number of corporate workers worldwide. This situation may, however, change but it depends on how quickly governments, employers, and training institutions respond with concrete action.
