More evidence suggests AI is reducing employment

An analysis by American financial researcher Eric Basmajian and data from a Goldman Sachs report point to a link between the spread of AI, rising productivity and falling employment in some US industries. Basmajian estimates that AI currently affects about 15% of jobs. The Financial Times notes that productivity figures do not indicate the quality or growth of output: they may reflect a reduction in headcount. This link is not apparent outside the US.

US data link AI adoption to rising productivity and unemployment

What the data show

Eric Basmajian studied industries that use AI most actively: information technology, professional services and finance. His data show that since 2022, productivity growth in these sectors has accelerated from 2.2% to 5%. Basmajian links this to a gradual increase in labor productivity amid the widespread adoption of AI.

The day after his analysis was published, a document began circulating on social media that was purportedly an internal Goldman Sachs report. It says AI’s impact on the labor market extends beyond the technology sector. That sector has already been significantly affected as generative AI has become a common tool for writing code.

The report points to a link between the extent to which an industry is exposed to AI and unemployment in non-technology professional fields. These include finance, telecommunications, technical services and administrative work. The data appear to align with Basmajian’s findings.

What limits the conclusions

The Financial Times offers several caveats about these data. The main one concerns how productivity is measured: the metric does not, by itself, reflect the quality of work. Higher productivity does not necessarily mean that AI has increased output or revenue. It may be linked to a reduction in the number of workers — a different outcome.

The data also cover only the US. The observed link between AI use, rising productivity and falling employment is not clearly apparent outside the country. The Financial Times suggests that the findings may also reflect broader economic difficulties that emerged after the pandemic, rather than the global displacement of workers by artificial intelligence.

Dataist's take: The new data strengthen the case that AI is affecting employment, but they do not yet prove that AI itself is causing job cuts. The conclusions are limited to the US, and productivity figures cannot distinguish increased output from a reduction in headcount.