The study released Wednesday by the Brookings Institution seems to contradict findings from previous studies — including Brookings’ own — that show lower-skilled workers will be most affected by robots and automation, which can involve AI. But it may be more accurate to say it complements those earlier studies. The new research is based on work by Stanford Ph.D. student Michael Webb, who developed a method to examine AI patents and compare the text with a Department of Labor database containing descriptions of occupational duties.
Brookings — which defined AI as “an increasingly powerful form of digital automation, based on machines that can learn, reason, and act for themselves” — used Webb’s data.
Webb found the most instances of paired verbs and nouns in AI patents and compared those with pairs found in occupational descriptions. From the patents, recognize was paired with nouns such as pattern, image, speech, face, voice, automobile, emotion, gesture and disease. Another verb, predict, was paired with the following nouns: quality, time, performance, fault, behavior, traffic, prognosis and treatment.
“Fully 740 out of the 769 occupational descriptions Michael Webb analyzed contain a capability pair match with AI patent language, meaning at least one or more of its tasks could potentially be exposed to, complemented by, or completed by AI,” the Brookings study said.
According to the study, some occupations that could be most affected by AI include market research analysts and sales managers, programmers, management analysts and engineers.
The lead author of the study, Brookings senior fellow Mark Muro, pointed out in an interview that its “analysis is a forecast of involvement with AI or exposure to it, but not necessarily worker displacement.” He expressed optimism that those with bachelor’s degrees or higher “will have greater skills and resiliency and ability to reorient their work than other workers.”
But in his study, Webb said older, higher-skilled workers could be negatively affected. “Older workers tend to be less mobile than younger workers, in terms of both occupation and geography, making it more difficult for them to adapt,” Webb said. “They also have fewer years of working life remaining, making educational and training investments less attractive for them.”
Both Muro and Webb were careful to stress that high exposure to AI isn’t necessarily negative, though. It could mean changes in duties but not necessarily the loss of jobs.
“The Bay Area will probably do pretty well,” Webb said in an interview. “It’s creating a lot of the underlying technology” for AI.
Patrick Kallerman, research director at Bay Area Council’s Economic Institute, agrees. He called the Bay Area unique and predicted that the region is unlikely to feel a “huge effect” from short-, medium- or even long-term automation.
But he added that “if research continues to show similar findings, then none of us (is) safe. It could shift the political-public policy landscape going forward.”
Other key findings by Brookings:
Because men are overrepresented in analytic technical and professional roles (plus production), they work in jobs with much higher AI exposure scores than women.
White and Asian American workers, who are overrepresented in technology, engineering, and legal-managerial occupations, will be affected most by AI.
People with bachelor’s degrees will be more than five times as exposed to AI than workers with a high school degree.
“Prime age” workers ages 25 to 54 — in professional and technical roles — have jobs that are going to be disproportionally affected by AI.
The metro area most likely to be affected by AI based on what kinds of jobs are there is the area that includes San Jose, Sunnyvale and Santa Clara. The area that includes San Francisco, Oakland and Hayward was No. 22 on the list.
Besides high-tech metro hubs, the study also found that areas most likely to be affected by AI are places heavy on agriculture and manufacturing, including Bakersfield and Salinas in California, and places like Detroit, Greenville, South Carolina, and Louisville, Kentucky.
©2019 The Cupertino Courier (San Jose, Calif.). Distributed by Tribune Content Agency, LLC.
