top of page

Reports of AI Ending Human Labor May be Greatly Exaggerated

12/10/23

Editorial team at Bits with Brains

Recent advances in artificial intelligence (AI), together with the uncertainties around Open AI, have renewed apprehension about the impact of technology on jobs.

Anxiety has accompanied previous waves of innovation, though history shows these fears were overblown. Now fresh evidence from Europe in the 2010s suggests occupations more exposed to AI-enabled technologies increased their employment share during the rise of deep learning.[1]


AI has enabled automation across a wide range of non-routine tasks in both manufacturing and services. Natural language processing, image recognition and other applications are automating human labor in virtually every occupational area. This general-purpose technology contrasts with earlier waves of computerization and industrial robotics that only automated a limited set of routine tasks.


Empirical evidence on AI's effects on jobs and wages is still evolving. Some U.S. studies find AI-exposed occupations experience no significant change in employment, though AI-adopting firms reduced hiring. This implies AI is replacing human labor in a subset of tasks. One study shows a small rise in wages for AI-exposed occupations. The literature has focused on the U.S.; new evidence examines 16 European countries from 2011-2019.


About 25% of European jobs were in occupations highly exposed to AI in the 2010s, based on two measures of exposure. One quantifies abilities needed for occupations that align with AI applications. Another measures the overlap between AI patent language and occupational job descriptions. Exposure level depends on whether AI substitutes or complements human labor.


The study finds a positive association between AI exposure and changes in European employment shares from 2011-2019. Moving up the distribution of AI exposure is associated with increased sector-occupation employment. AI exposure is linked to bigger employment gains for high-skill occupations and younger workers, unlike previous computerization. No evidence was found of the polarization seen with software automation.


AI's positive employment association holds for most European countries, though impact size varies. This may reflect differences in technology diffusion, education, regulation, and competition. AI exposure shows no significant effect on wages, except slightly slower wage growth in only one measure.


It's too soon to reach a verdict on AI's impact as its technologies continue advancing. Generative AI may well have changed everything and much of the effect on jobs and wages has yet to materialize. But the study offe


The future remains uncertain whether AI will ultimately substitute or complement human labor across the economy. Skill levels, demographics, diffusion rates and policy choices will help determine who are the winners and losers from AI adoption.


Sources:

[1]European Central Bank, 28 NOVEMBER 2023 ยท RESEARCH BULLETIN NO. 113

Sources

bottom of page