While genetic AI like ChatGPT represents a major advance, it’s certainly not the first time a new tool is poised to change (or even eliminate) certain jobs. And this past experience can offer important lessons about what we can expect in the future.
In a new paper, Kellogg’s Dimitris Papanikolaou and Bryan Seegmiller (professor and assistant professor of economics, respectively) take on the question of how the arrival of new technology affects workers and their earnings. The study was authored by Leonid Kogan and Lawrence Schmidt of the MIT Sloan School of Management.
Their analysis focuses on the period from 1981 to 2016—a time period in which many new technologies were introduced to a variety of occupations, both blue-collar and white-collar.
Perhaps not surprisingly, researchers have found that when a new tool can perform a task in his place one worker, all affected workers suffer. “They’re experiencing wage loss, and that’s largely independent of their age, their income level, the field they work in, the type of work they do, or whether they have a college degree,” Papanikolaou says. But when a new technology completes For workers performing a task, the results are more variable: more experienced and highly paid workers suffer, while new hires appear to benefit.
Papanikolaou, Seegmiller and their colleagues also studied the potential effects of artificial intelligence on today’s workers. They found that “artificial intelligence, as a technology, is leveling the playing field in a profession,” Papanikolaou says. In other words, if everyone can code, a skilled and experienced coder will be less valuable in the job market. The result is that “it will hurt the workers who are best at their jobs.”
Measuring exposure to new technology
The researchers began by collecting job descriptions from ONET, the Occupational Information Network, as well as the 1991 edition of the Dictionary of Occupational Titles—both widely used sources of information about different occupations and the functions they perform. For example, the ONET listing for the position “Preschool Teacher, Non-Special Education” lists 37 tasks, including “Demonstrate activities to children” and “Read books to whole classes or small groups.”
Then—in a perhaps ironic twist—the researchers asked ChatGPT to classify the tasks of each job as either routine (requiring little experience and likely easy to automate) or non-routine (requiring a lot of experience and likely difficult to automate). . They also validated the results of ChatGPT against other classification methods to ensure their accuracy.
For example, as a professor, “sometimes I teach, sometimes I write papers, sometimes I do reimbursements,” explains Papanikolaou. Dealing with returns “is rather routine, while the other two are completely non-routine.”
They then compiled a list of important patents issued from 1980 to 2007. They focused on so-called breakthrough patents—those that were very different from previous patents but had a big influence on future patents.
The team then calculated how closely these breakthrough patents corresponded to routine and non-routine tasks. If a patented technology was closely related to a routine task, researchers considered it laboriouseconomy technology—one that is likely to fully automate this task. If a patented technology was closely related to an unusual task, researchers considered it to be laborincreasing technology—that is likely to complement a worker performing this task.
This information was used to calculate measures of exposure to labor-saving and labor-enhancing technology for different occupations. A job experienced high exposure to labor-saving technology if many of its tasks were automated within a given time period. Similarly, a job experienced high exposure to work-enhancing technology if many of its non-routine tasks were supplemented by technology within a given time period.
How exposure to technology affects occupations and workers
To determine the consequences of exposure to labor-saving and labor-increasing technologies, the researchers collected US government data on workers in different occupations, including their earnings, ages, and education levels from the years after the pioneering patents were granted .
Overall, for any occupation, exposure to labor-saving technology predicted lower wages and lower employment. Exposure to job-enhancing technology, meanwhile, predicted higher wages and higher employment for that job.
But when researchers shifted gears and began analyzing the effects of technology exposure on worker level, despite occupation
level, they discovered a more complicated story—particularly when it came to work-augmenting technology.
Across all occupations, the average worker whose exposure to job-enhancing technology suddenly increased saw a small decrease in earnings and a small increase in the likelihood of losing their job. These trends were even more pronounced among white-collar workers, older workers, and highly paid workers in an occupation.
That finding, combined with the knowledge that wages rose at the professional level, “leads you to suggest that a lot of the benefits are going to newly hired workers,” Papanikolaou says. In other words, workers who were used to doing things a certain way found it difficult to adapt when complementary technology arrived, while less experienced workers were able to take advantage of the power of these new tools.
What does the future of artificial intelligence hold?
But will the trends of technological progress of the past decades extend to artificial intelligence? To find out, Papanikolaou, Seegmiller and their colleagues did a new analysis. This time, instead of using patents as a proxy for technological change, the researchers asked ChatGPT whether AI could perform a given task without human intervention, or whether the task would require significant human intervention.
As before, the researchers combined this information with job descriptions to determine whether a job’s routine and non-routine tasks could be supplemented or replaced by artificial intelligence. They used census data on Americans’ earnings to predict the effect of AI exposure on wages.
Although speculative, the findings suggest that workers in office and administrative occupations, as well as those in manufacturing and transportation, face high levels of exposure to AI as a labor-saving technology.
Meanwhile, the occupations most exposed to AI as a complementary technology included “Insurance Underwriter,” “Medical Transcriptionist,” “Customer Service Representative,” “Personal Financial Advisor,” and “Budget Analyst.” While less experienced workers in these roles may benefit from robotic assistance, the researchers suggest, older ones may struggle and see their wages decrease because their expertise and experience may not provide the competitive edge that once had
Compound effects
Papanikolaou says the research offers a much-needed perspective on how artificial intelligence—or any new technology—might change the workforce.
“Especially with AI, everyone freaks out and interprets the statement ‘AI will affect my job’ as ‘AI will do my job for me.’ And two things are not the same, because artificial intelligence can be a tool or a substitute,” he says. “And just because the job can change doesn’t mean the job will be eliminated.”
The findings also suggest that soft skills may be becoming increasingly important in the workforce. In fact, jobs that rely primarily on interpersonal skills have barely been affected by technology.
For example, Papanikolaou points out, AI tools may, in the future, help doctors and nurses make medical diagnoses — but they likely won’t be able to provide the emotional care and support sick patients need: “This it’s something AI will probably do much worse.”
