What the RBA observes
An analysis by Mike Major for the Reserve Bank of Australia (RBA) built a monitoring system focused on the segments where the effects of artificial intelligence should emerge first: early-career workers and the most exposed occupations, such as programming, office work and customer service. According to The Decoder, since ChatGPT’s launch in late 2022 job postings have fallen across all groups, but the sharpest contraction was recorded in the most exposed group. Within that same group, 15–24-year-olds and young graduates show weaker employment rates and higher unemployment than older workers.
The finding should be read with caution: the study does not establish a direct causal link and measures only theoretical exposure, not companies’ actual use of AI. Major himself warns that changes in the economic cycle or in remote work after the pandemic could explain the observed figures.
International comparisons and new questions
The Australian picture is not isolated. The Decoder cites an August 2025 Stanford study based on millions of U.S. paychecks, which found a relative 13 percent reduction in employment among 22–25-year-olds in exposed occupations from autumn 2022, and a nearly 20 percent reduction among young software developers; more experienced workers reportedly remained stable or grew. A March 2026 Anthropic analysis reportedly indicated a decline in hiring rates of about 14 percent in the same age group, though without a systematic increase in overall unemployment.
The company itself offers a corporate case study: according to an internal review, the share of junior hires in technical roles fell from 27 to 9 percent since the start of 2025. Anthropic admits that it is a special case, given its intensive use of AI. Researchers speak of a possible “tragedy of the cognitive commons”: the disappearance of junior roles could create a shortage of experienced professionals in the future—assuming they are still needed.
Macroeconomic effects still uncertain
Other studies downplay the scale of the phenomenon. A large study of 25,000 Danish workers, with an AI adoption rate of 64 percent, found no measurable changes in wages or hours worked, with average time savings of just 2.8 percent. However, a January 2026 analysis of 10.6 million LinkedIn profiles and U.S. unemployment data, led by Morgan Frank of the University of Pittsburgh, places the onset of difficulties in exposed occupations in the first months of 2022, before ChatGPT: macroeconomic factors, such as the Fed’s change in interest-rate policy or a correction after pandemic hiring, could weigh more heavily than previously thought.
At the same time, AI is creating new demand: the World Economic Forum’s Future of Jobs 2025 report indicates that 77 percent of surveyed companies prioritize reskilling over cuts and forecasts a net gain of 78 million jobs by 2030, while LinkedIn reports a sixfold increase in AI-related job postings in a year.
