New research is challenging fears that artificial intelligence technology will displace a large portion of the world’s workforce, with data from Europe finding occupations exposed to AI over the past 20 years actually saw an increase in employment.

Stefania Albanesi, one of the authors of a report published Tuesday by the European Central Bank, noted that while the data referenced only covers AI advances up until 2019, it demonstrates how the technology has interacted with the labour market so far.

“The study covers the 2000s and 2010s, where the kind of AI that was implemented was basically deep learning and machine learning models,” Albanesi, an economics professor at the University of Miami, told BNN Bloomberg in television interview.

That form of AI is different from large language models such as ChatGPT deployed in the last year, she noted, but the data is still an informative look at the possible future of work.


The report found that in the “deep learning boom of the 2010s, occupations potentially more exposed to AI-enabled technologies actually increased their employment share in Europe.”

The study of 16 European countries found that the jobs created were primarily for younger, high-skill workers – mainly college graduates.

Albanesi said that these findings are a break in a pattern from the last major technological advance – the launching and diffusion of the internet – which caused a decline in routine administration and production jobs.

“Instead, for this wave, which was associated with the early phase of expansion in AI in the last 20 years, we saw a growth in employment.”

The report found that for occupations requiring “low and medium-skill” workers, AI exposure didn’t have a significant effect on employment numbers.

But for occupations requiring high-skill workers, it found “a positive and significant association.” AI exposure appeared to boost employment share by 3.1 per cent using one measure cited in the report, and 6.7 per cent using another measure.


The researchers said that historically, technological advances have been accompanied by fears about potential job losses.

“This apprehension persists, even though history suggests that previous fears about labour becoming redundant were exaggerated,” the report said.

Albanesi said that until recently, AI technology has been limited to replacing routine and repetitive labour tasks.

However, she cautioned that new large language models could potentially replace non-routine jobs, which make up a larger portion of the workforce.

“That is sort of the risk,” she said.

“But that does not mean that in the aggregate, there will be a net decline in available jobs because new jobs may be generated in response to these abilities we are granted because we have access to this technology.”