The next wave of laborsaving technology will wear a white collar. Artificial intelligence—programming machines to think more like humans—is expected to alter how Americans work on a scale similar to the impact of robotics and desktop computers. But employees affected by it are much more likely to be in management or professional roles than laborers turning screws or filing papers.
The most vulnerable occupations include marketing specialists, financial advisers and computer programmers— jobs that tend to pay high wages and skew toward male, white and Asian workers, a recent study from the Brookings Institution found. Other jobs most vulnerable to being affected by AI included certain types of engineers, optometrists, graphic designers, software developers and sales managers. New technology in the workplace has generally been better for higher-skilled workers than for the lowerskilled, said Mark Muro, one of the study’s authors. “Artificial intelligence could play out just the opposite.” While machines have long been able to perform repetitive physical tasks or complex mathematical calculations, AI enables computers to analyze data, predict outcomes, learn from experience by recognizing patterns and make decisions. Such tasks are curworkers rently done by professional workers, many with college degrees.
Think of how a financial adviser analyzes a client’s economic circumstances, income prospects and personal goals to provide guidance on retirement planning. With AI, an algorithm could use the same information to make recommendations. Another example would be the market-research analyst who analyzes consumer spending trends and patterns to recommend an advertising strategy for a new movie or automobile. With AI, a computer could do the same thing.
Similarly, medical facilities have started using computers to read X-rays and determine whether images are consistent with a disease such as pneumonia, a task previously performed by radiologists. The technology could free up medical personnel to spend more time with patients or examine less clearcut cases—and it could reduce the overall need for radiologists. “AI will substitute for a set of tasks, but there’s no reason it would have to be a total displacement,” said Stanford University economist Michael Webb. “The only thing you can say for sure is that the job will change.” In contrast, automation has replaced people securing a bolt on an assembly line, adding figures on a ledger, or, more recently, flipping a burger.
To study the possible effects of AI, Mr. Webb employed it to review more than 16,000 patents related to AI to determine capabilities, such as “diagnose disease” or “recognize aircraft.” Meanwhile, he examined a Labor Department database of hundreds of occupations to catalog specific tasks required for jobs. He then matched how frequently the potential AI abilities overlapped with what is required at existing jobs to develop a measure of exposure. Mr. Webb’s analysis provided the basis for the Brookings study.
The study found holders of bachelor’s degrees are five times as likely to be exposed to some effect of artificial intelligence as those with just high-school diplomas. Those with the highest exposure are in the 70th to 90th percentile of wage earners. It is possible artificial intelligence will allow some to dispense with time-consuming tasks such as data analysis, and focus on potentially more profitable activities, such as meeting clients. Those workers could become more productive and command higher wages. Other workers could find their jobs simplified and more easily filled by someone with less education, which could drive down wages in the process. And in other cases, jobs could be replaced entirely by technology.
To be sure, experts in the past have incorrectly predicted that other major shifts in the economy—including the mechanization of agriculture, automation of factories and outsourcing of labor to foreign countries—would cause mass unemployment. While these trends did eliminate many U.S. jobs, many others were created over time. And the unemployment rate was near a 50-year low in January, at 3.6%. Although the U.S. has six million fewer factory jobs than it did in the late 1970s, it has 62 million more total jobs, primarily due to the growth of the service sector. Top executives are likely less exposed to potential displacement by AI than lower level professionals. The Brookings study found that exposure to artificial intelligence falls for the top 10% of earners. “CEOs are largely exempt from this,“Mr. Muro said. “They’re not the ones doing the number crunching and making the PowerPoints.