Join getAbstract to access the summary!

What Jobs Are Affected by AI?

Join getAbstract to access the summary!

What Jobs Are Affected by AI?

Better-paid, better-educated workers face the most exposure

Brookings Institution,

5 min read
5 take-aways
Audio & text

What's inside?

New research suggests unexpected effects of artificial intelligence on the labor market.



Editorial Rating

10

Qualities

  • Innovative
  • Concrete Examples
  • Hot Topic

Recommendation

Artificial intelligence (AI) will transform the workplace; the question is how. Until now, estimates of AI’s effects have been commingled with analyses of automation and software impacts. Stanford researcher Michael Webb has found a way to coax out granular answers regarding the potential effects of AI on particular occupations. The method yields scores for exposure level to impact from AI on specific jobs, roles, demographic groups and communities. Senior fellow Mark Muro, research analyst Jacob Whiton and research associate Robert Maxim of the Metropolitan Policy Program at the Brookings Institution report on the Stanford findings in their fascinating white paper.

Take-Aways

  • A researcher has used a new approach to predict the effects of AI on the labor market.
  • AI will affect the workforce in different ways than it affects robotics and software.
  • AI will exert strong effects on the business, finance and technology industries.
  • Men, midcareer workers, and white and Asian-American workers will likely see heavy impacts from AI.
  • AI’s impact will vary geographically depending on local variations in industries, occupations and educational levels.

Summary

A researcher has used a new approach to predict the effects of AI on the labor market.

Michael Webb of Stanford University has devised a novel statistical method of assigning exposure scores to occupations. The data, when combined with other workforce information, make possible granular predictions of AI’s future impact on labor markets. Webb used AI to compare the tasks mentioned in AI patents to the US Department of Labor’s Occupational Information Network (O*NET) database occupational descriptions. Where these overlap, Webb predicts that AI has potential to affect the occupation. A high exposure score indicates a likelihood of AI exerting either positive or negative impact. The method can’t predict whether, or in what way, that potential will play out. For example, AI could benefit an occupational group by automating certain tasks, thereby giving workers the possibility of taking on richer roles. And AI could generate new kinds of jobs, just as the invention of the automobile did.

“What’s coming may not resemble what we have been experiencing or expect to experience.”

AI will affect the workforce in different ways than it affects robotics and software.

In previous analyses that studied impacts, researchers wrapped together AI, automation and software. Analyzing AI in isolation gives different results. AI does share with robotics and software the fact that its impact will pervade nearly all occupational groups. But AI’s projected pattern of greatest disruption in labor reverses that of robotics and software. AI will primarily affect workers with higher educations and salaries. Managers, supervisors, analysts, programmers and engineers will see significant effects from AI. Service workers will see a lower level of impacts. Although AI will affect a broad swath of jobs, only 18% of workers will face high exposure. For 48% of jobs, AI will exert low or no impact.

Geographically, AI’s impacts resemble those of automation, except that Nevada has high exposure to automation but low exposure to AI. Washington State flips in the other direction, with the state seeing high exposure to AI impacts although its exposure to automation impacts remains low. Exposure to AI effects by type of community reverses the pattern of automation: AI’s impacts will affect larger urban communities more than smaller rural communities.

“The substitution of AI for some well-paid human prediction work is a certainty.”

AI will exert strong effects on the business, finance and technology industries.

Automobile manufacturing, the textile industry, software publishing and computer system design will see strong impacts from AI. Health care, education, retail, and food and beverage will experience low impacts. Unlike automation, AI can perform nonroutine work. For example, AI can already identify defects in production processes, sense possible equipment failures and optimize procurement processes. AI could also affect law practices, eliminating much of the labor that paralegals and researchers currently perform while providing faster and more accurate results.

Men, midcareer workers, and white and Asian-American workers will likely see heavy impacts from AI.

Gender differences occur because men participate at higher levels than women in AI-exposed roles and occupations. Because women tend to find employment in education, health care and personal care, they will collectively see lower impacts from AI. Age differences exist because young workers tend to populate low-skill jobs – which score low on AI impact exposure – while midcareer workers tend to occupy managerial, analytical and technical roles. The same root cause accounts for higher AI impact exposures for white and Asian-American workers than black or Latino/Hispanic workers. The former groups tend to work in technical and managerial roles, while members of the latter groups more often find employment in service occupations – which will see lower impacts from AI.

“The most elite workers – such as CEOs – appear to be somewhat protected.”

AI’s impact will vary geographically depending on local variations in industries, occupations and educational levels.

AI will heavily affect US manufacturing regions, as well as the managerial/high-tech US coasts. Among metro areas, the large cities with highest exposure to AI impacts include San Jose, Seattle and Salt Lake City. Bakersfield, California, scores highly because it serves as a logistics hub, and Boulder, Colorado, faces high exposure resulting from its high-tech sector. Cities whose economies rely mainly on services, such as Las Vegas and El Paso, score lower in exposure to AI impacts.

About the Author

Mark Muro is a senior fellow, Jacob Whiton is a research analyst and Robert Maxim is a research associate in the Metropolitan Policy Program at the Brookings Institution.

This document is restricted to personal use only.

Did you like this summary?

Read the article

Comment on this summary