The rise of generative AI: modelling exposure, substitution and inequality effects on the US labour market

Summary

Focus

We investigate the evolving impact of artificial intelligence (AI) on the US labour market, examining how AI might complement or substitute human labour across 711 occupational categories. We model the relationship between cognitive AI capabilities and the complexity of job-related skills, highlighting different effects on core and side skills and investigating AI's impact across the wage distribution.

Contribution

We incorporate how various occupations' exposure to AI depends on the difficulty of the skills they require and whether these skills are within reach of the AI's technological capabilities. We also model the impact on side versus core occupational skills: for example, an AI capable of bookkeeping helps doctors with administrative work, freeing up time for medical examinations, and may thus increase their productivity. But it risks the jobs of bookkeepers.

Findings

Although high-wage occupations are more exposed to AI, low-wage occupations are more at risk of being replaced by it. At high AI capabilities, up to 45% of skills in the highest wage quartile are exposed to AI, compared with only 26% in the lowest wage quartile. The lower overall exposure for low-wage occupations is not due to the difficulty of the required skills, but because they require interactions in the physical world, eg changing a hospital bed.

Even though high-wage occupations show higher overall exposure, AI usually complements them by handling some of their side skills, while their core skills remain beyond AI's capabilities. In contrast, low-wage occupations are less exposed, but if they are exposed, AI can typically perform their core skills. For example, a high-wage lawyer might use AI to help with legal research (a side skill), while a low-wage occupation in the same profession, such as a legal clerk, could be substituted by AI tools that manage their core tasks of scheduling and document preparation. On balance, this makes low-wage occupations more likely to be substituted by AI.

This may exacerbate economic inequalities. Despite a common belief that AI could displace many white-collar roles, these positions appear less vulnerable in the medium term due to the complexity of their core skills. We also discuss policy implications in the light of these findings, stressing the importance of skill development, AI transparency and international cooperation in labour policies to mitigate adverse impacts.

Abstract

How exposed is the labour market to ever-advancing AI capabilities, to what extent does this substitute human labour, and how will it affect inequality? We address these questions in a simulation of 711 US occupations classified by the importance and level of cognitive skills. We base our simulations on the notion that AI can only perform skills that are within its capabilities and involve computer interaction. At low AI capabilities, 7% of skills are exposed to AI uniformly across the wage spectrum. At moderate and high AI capabilities, 17% and 36% of skills are exposed on average, and up to 45% in the highest wage quartile. Examining complementary versus substitution, we model the impact on side versus core occupational skills. For example, AI capable of bookkeeping helps doctors with administrative work, freeing up time for medical examinations, but risks the jobs of bookkeepers. We find that low AI capabilities complement all workers, as side skills are simpler than core skills. However, as AI capabilities advance, core skills in lower-wage jobs become exposed, threatening substitution and increased inequality. In contrast to the intuitive notion that the rise of AI may harm white-collar workers, we find that those remain safe longer as their core skills are hard to automate.

JEL Classification: E24, E51, G21, G28, J23, J24, M48, O30, O33

Keywords: labour market, artificial intelligence, employment, inequality, automation, ChatGPT, GPT, LLM, wage, technology