Tech Explained: Is AI Turning Back the Clock on the Job Market? in Simple Terms

But AI seems poised to follow the opposite trend. The ability of AI to handle office tasks, such as copywriting and data analysis, seems more likely to harm white-collar jobs than blue-collar jobs, where hands-on skills remain paramount.

In some sense, “AI is turning back the clock,” says Bryan Seegmiller, a Kellogg assistant professor of finance.

Research by Seegmiller and his colleagues supports this prediction. They used large language models to analyze nearly 200 years of data on technological innovations and occupations. And they found that, as the use of AI progresses in the workplace, relative demand for white-collar jobs declines, with blue-collar jobs picking up a larger share of workers.

The team included Huben Liu, a Kellogg PhD student in finance; Dimitris Papanikolaou, a Kellogg professor of finance; and Lawrence Schmidt at the MIT Sloan School of Management.

The researchers’ findings extend a more-recent pattern, starting in the late 20th century, where technology has increasingly replaced cognitive tasks common in white-collar professions. “This is a 50-year trend in the making,” Seegmiller says.

That doesn’t mean college students should immediately drop out of school and train to become plumbers. But they should zero in on skills that AI can’t yet replace and take advantage of AI to perform better in other parts of their job, Seegmiller adds.

“[You need to] have a healthy appreciation of, and respect for, technological forces,” he says, “and to understand the forces that are working in your favor and that are working against you.”

Learning from the past

Seegmiller and his collaborators previously found that, when AI directly replaced most of the tasks for jobs held from 2014 to 2023, it reduced demand for human workers in that field. But if AI took over only a subset of a profession’s tasks, the effect of the technology on employment was actually positive.

The researchers wondered if this pattern held more broadly across other technologies throughout history. If so, it could provide a frame of reference to predict what might happen to the labor market as AI improves.

“We always want to know how things are going to unfold,” Seegmiller says. “The only thing we have the ability to learn from is the past.”

For the current study, the team gathered data on technological patents from 1850 to 2024, covering everything from steel-production methods to e-commerce systems. They also used a large language model to generate descriptions of tasks for jobs listed in the U.S. Census during the same period.

Then they compared the text from patents with the descriptions of job tasks in each decade using natural language processing techniques that they had developed in previous research on the impact of technology on the labor market. Whenever the description of a patented technology and a job task closely aligned, that task was considered “exposed” to the technology. In other words, the technology had the potential to replace a human worker for that task.

As one might expect, manual tasks were by far the most highly exposed to technology from the mid-19th century onward. Then, around the mid-20th century, an important shift occurred: thanks to advancements in computers and IT, cognitive tasks steadily grew more exposed to technology, though manual tasks still remained the most exposed overall.

In an analysis from 1910 onward, the researchers found that greater exposure to technology generally hurt a job’s employment growth. However, if the exposure was limited to a small subset of tasks, it actually benefitted employment growth for a given job because it allowed workers to focus their attention on other tasks.

In all, the results confirmed that the historical forces driving the effects of technological change on jobs have important parallels with how AI is likely to impact labor markets. As Seegmiller notes, “While the capabilities of AI may feel unprecedented, it’s similar to previous technologies in some important ways: the ability to reallocate effort toward remaining unexposed tasks within a job always helps reduce the negative impact, and interpersonal skills always appear the least exposed.”

A reversal of trends

The team then calculated the net effect of technology exposure and innovation on different groups of occupation types.

They found that, throughout the 20th century, higher-paying jobs generally benefited from technology—as did occupations that required more education and jobs that employed more women, such as clerical and administrative jobs. In contrast, lower-paying “middle-skilled” trade jobs, such as those in transportation and manufacturing, which typically employed more men, suffered.

However, when the researchers used a mathematical model to simulate how AI would affect different occupations over the next 5 to 10 years, the picture that emerged aligned with a newer trajectory. It predicted that AI would decrease the relative demand for white-collar jobs. In that sense, their model expects “a reversal of previous trends,” Seegmiller says.

To be specific, it projected that demand would fall for jobs requiring a high level of education compared with jobs requiring less education—while demand for jobs requiring an average level of education would fall the most. It also projected a drop in demand for traditionally higher-paying jobs, such as clerical, technical, and managerial positions. Finally, demand for jobs that tend to have a higher share of women would drop compared with jobs with a higher share of men.

That’s not to say that the total number of blue-collar jobs will necessarily increase. Rather, this prediction suggests that, of the jobs that remain in a post-AI world, a larger fraction of them will likely fall into that category.

This AI-driven reversal of past trends reflects an essential difference between AI and other technological changes in the past. “Whereas earlier technologies always exposed a mix of manual and cognitive work, AI technologies really laser in on white-collar, cognitive-focused jobs,” Seegmiller says.

No easy answers

While the response of some white-collar workers might be to consider switching to a profession that is less exposed to AI, Seegmiller advises that people approach the study’s findings with care.

The model’s predictions are like “painting with a broad brush,” he says. And even within occupations hit hard by AI, “there are ways to differentiate yourself.”

For example, as AI automates some tasks, workers could focus on building skills that require a human touch, such as interpersonal communication, collaboration, and creative thinking. Throughout the study period, very few interpersonal tasks have been displaced by technology, and Seegmiller believes that will generally continue to be the case, “even if ChatGPT might be able to pretend to be a therapist.”

Workers can also boost their productivity by using AI to perform simpler tasks for them. That might allow workers to “focus on the part where our added value is,” he says. “Maybe that’s using these tools to become more productive at generating ideas or seeing the big picture.”

Still, the researchers acknowledge that there are no easy answers for how workers should cope with the seismic changes ahead.

“There’s no golden ticket here,” Seegmiller says. But by carefully considering how technology can both hurt and help, you can “make informed decisions about the way you invest in yourself.”