Clarity for AI buildersSponsored by AI Routing
adoption

RBA analysis finds weaker employment among young workers in AI-exposed jobs

RBA analysis finds weaker employment among young workers in AI-exposed jobs

An RBA analysis finds weaker employment among young workers in AI-exposed jobs, but does not establish that AI is the cause.

According to The Decoder’s account of research by Mike Major for the Reserve Bank of Australia, employment among young people in occupations with the highest potential exposure to AI has declined since late 2022, while employment among older workers in those occupations has risen. The report says there is little evidence of widespread AI-driven job displacement in Australia so far.

Major’s monitoring approach focuses on groups where changes might emerge first: entry-level workers and people in highly exposed occupations, including programming, clerical work and customer service. Job postings have fallen across all exposure groups since ChatGPT launched in late 2022, according to the account, with the steepest decline in the most exposed group.

Within that highest-exposure group, workers aged 15 to 24 and young university graduates have weaker employment rates and higher unemployment than older workers. That difference makes the entry-level labor market an important part of the findings: an absence of widespread displacement can coexist with worsening prospects for people trying to enter particular occupations.

Exposure is not evidence of adoption

The central limitation is that the analysis measures theoretical exposure to AI, not companies’ actual use of the technology. Major cautions that the findings do not establish direct causation. Post-pandemic business-cycle changes and shifts in remote work could also help explain the employment patterns.

The distinction matters because an occupation’s tasks being potentially affected by AI does not show that employers have automated those tasks or reduced hiring as a result. The observed gap between younger and older workers is a signal to investigate, rather than a measure of jobs lost to AI.

The timing also leaves room for competing explanations. The Decoder cites a January 2026 analysis of 10.6 million LinkedIn profiles and US unemployment data, led by Morgan Frank at the University of Pittsburgh, which found that difficulties in exposed occupations began in early 2022—before ChatGPT’s launch. That research points to interest-rate changes and a correction after pandemic-era hiring surges as possible contributors.

Other studies show a similar age divide, with limits

The Decoder also reports that an August 2025 Stanford study using millions of US payroll records found a 13 percent relative decline in employment among 22- to 25-year-olds in AI-exposed occupations since late 2022. The decline was nearly 20 percent among young software developers, while more experienced workers in the same occupations held steady or gained ground.

A March 2026 Anthropic study, as described by The Decoder, found job-finding rates down about 14 percent among 22- to 25-year-olds in exposed occupations, but no systematic increase in overall unemployment. These findings concern different countries, age ranges and labor-market measures; they provide context for the Australian pattern, not direct confirmation of its cause.

Broader outcomes remain mixed. The supplied report describes a study of 25,000 workers in Denmark that found no measurable change in wages or hours worked despite a 64 percent AI adoption rate. Average time savings amounted to 2.8 percent of working time. It also cites SEEK data indicating that AI is complementing existing work and generating demand for AI skills.

Taken together, the evidence supports a narrower conclusion than a general claim of AI-driven job losses: young workers in some highly exposed occupations are facing weaker outcomes, while economy-wide effects remain difficult to establish. The practical question is whether reduced entry-level opportunities persist as actual AI adoption grows, and how much of that change can be separated from the wider hiring cycle.

This draft is based on The Decoder’s supplied account; the underlying RBA analysis and other cited studies were not supplied for independent examination.