A new report by the Financial Times in partnership with Focaldata suggests that artificial intelligence is not flattening the workplace as many expected. If anything, it is reinforcing the very divides it was supposed to disrupt.
Based on a survey of 4,000 workers across the US and UK, the findings show a sharp imbalance in who is actually using AI. More than 60 per cent of high earners report using AI tools daily, compared with just 16 per cent of lower-income workers. The implication is straightforward: those already ahead are gaining even more leverage from the technology.
According to Daron Acemoglu, this outcome challenges the popular narrative around AI as a democratising force. He notes that effective use of these tools often requires a level of education, technical familiarity and abstract thinking that not all workers have access to. In his words, “the rhetoric out there is that the tools are going to be democratising,” but in reality, the barrier to meaningful use remains high. He adds a more direct warning: “AI is going to increase inequality between labour and capital… it is setting us up for a… shitshow.”
What is perhaps more surprising is who is adopting AI fastest. The data suggests it is not the youngest workers, but those in their thirties with more experience and longer tenure. Ronni Chatterji says this aligns with what the company is seeing internally, where AI tends to complement existing expertise rather than replace it. In practice, this means experienced workers are using AI to amplify their output, not level the field for newcomers.
The divide is not only about income or experience. The report also points to a persistent gender gap, with men significantly more likely to use AI tools across industries. Fabien Curto Millet estimates that women are about 20 per cent less likely to use AI than men. However, he also points to evidence that this gap can be narrowed. A 2025 training intervention, for example, led to a tripling of daily AI usage among women over 55, suggesting that exposure and structured learning play a critical role in adoption.
Across professions, the pattern is consistent. Workers in high-skill, white-collar roles such as law, accounting and software development are far more likely to integrate AI into their daily workflows than those in lower-paid roles within the same sectors. Chris Pissarides captures this shift succinctly, arguing that as technology becomes more advanced, the value of individual intelligence and skill increases alongside it.
Beneath these trends is a more structural concern about the future of work. If senior employees can use AI to perform tasks that were once handled by junior staff, the traditional pathway for building experience may begin to erode. That raises questions about how new entrants to the workforce will develop the skills needed to advance, especially in industries where early-career roles serve as the foundation for long-term growth.
There is some historical precedent for this kind of imbalance. Carl Benedikt Frey notes that similar disparities emerged during the early days of personal computing, before eventually evening out as the technology became more widespread. Still, he cautions that the timeline matters. If the gap takes a decade or more to close, the social and economic consequences could be significant.
David Adubiina
