How AI could make workers more productive – but paid less

The gains in worker productivity enabled by artificial intelligence (AI) are concentrated at the lower end of the skill and income spectrum, and economists and trade unions believe this phenomenon will push jobs to lower-wage regions of the world. It warns that this could accelerate outsourcing practices.

Several recent studies have shown that introducing language-based AI software into the workplace benefits lower-skilled workers more than higher-skilled workers, resulting in lower-skilled workers becoming less valuable. This suggests that there is potential to attract higher-paying jobs overseas.

“Especially in the case of generative AI that supports low-skilled workers, companies have the opportunity to replace more skilled workers with lower-skilled, lower-wage workers,” said Karl Benedict Frey, an economist at the University of Oxford. “I can do that,” he told The Hill.

“Maybe [lead to] “There is a big wave of offshoring as companies begin to tap into large pools of cheap labor in regions with high wage disparities, such as India, Bangladesh and the Philippines,” he said.

AI programs like ChatGPT not only increase the overall productivity of workers, but also reduce the gap between better and less productive workers, essentially homogenizing the workforce .

A Massachusetts Institute of Technology (MIT) study led by Shaked Noi and Whitney Chan and published last year found that “ChatGPT compresses the productivity distribution by providing more benefits to lower-skilled workers. “As a result, inequality among workers is reduced.”

Another 2023 study from Stanford University and the Massachusetts Institute of Technology found that generative AI resulted in an average 14 percent increase in general productivity, but a 34 percent increase among less experienced workers . This study by Erik Brynjolfsson and Lindsey Raymond looked at how AI is helping improve productivity and employee experience, particularly in outsourcing environments such as call centers.

The authors say their results are consistent with other studies, all of which show that generative AI systems[compressed] The distribution of productivity widens, with lower-skilled workers benefiting the most. ”

Policymakers are already sounding the alarm about the long-term impact AI could have on the labor market.

International Monetary Fund (IMF) Director Kristalina Georgieva said on Monday that AI was hitting the global labor market “like a tsunami”, Reuters reported.

He said AI could impact 60% of jobs in developed countries and 40% of the global job market within the next two years.

U.S. labor unions are increasingly concerned about the pace of adoption of this technology, the extent of its integration into production processes, and the fact that company leaders are making these changes without consulting employees.

The AFL-CIO, the nation’s largest labor union, said in a recent statement that “employers now use increasingly sophisticated tools to hire, monitor, evaluate, discipline, and even fire workers without their consent.” “We are introducing AI-enhancing technology that will transform the system.”

After last year’s strike, entertainment officials at the Writers Guild of America won assurances from studios that AI would not be allowed to write any literary work and that any AI-generated work would not be considered original work.

However, such guarantees may be eroded as consumption habits increasingly shift to social media and other online platforms that do not maintain labor contracts with content creators.

The standardization of productivity brought about by AI, and the resulting opportunities for companies to reduce overhead and outsource labor, is contributing to a phenomenon known as the productivity-wage gap that has persisted since the beginning of commercial computing. may worsen.

Fears that computers would cause mass unemployment turned out to be unfounded, but there was a gap between what workers produced and the wages they received. It’s getting bigger It started in the 1970s, but really took off in the 1990s and 2000s.

American workers have increased their hourly output by about 140 percent since 1973, but inflation-adjusted employee compensation has increased by only 45 percent over the same period. This means that workers receive far less wages compared to how much value they bring to their employers. The size of the gap depends on the measure of inflation used, but the gap persists no matter how you calculate it.

Over roughly the same period, the share of the economy that goes toward compensating labor, rather than returning profits to shareholders, fell from 64% of gross domestic product (GDP) to 59% in 2019. Meanwhile, the percentage of profits doubled. , increasing from 4 percent of GDP in 1974 to 8.5 percent in 2022.

During this time, major changes in U.S. tax law have been enacted, along with new labor and trade practices associated with globalization, but cross-border production coordination made possible by the widespread adoption of technology has also increased productivity. This is a contributing factor to the wage gap.

“It was beneficial to be at the forefront of the computer revolution.” [to the United States] Oxford University’s Frey attributed the split to increased offshoring and more aggressive automation of production processes, noting that the wedge between productivity and wage growth began in the 1980s.

As for what jobs could ultimately disappear as a result of generative AI, Frey said the jobs most likely to be automated are repetitive tasks that occur in structured environments. I have listed the jobs that include. This includes transportation, logistics, and warehousing operations, with much of the labor occurring in highly controlled processes.

Jobs that could see increased wage stagnation as a result of outsourcing include software design, computer programming, and content production, such as in the animation field, where countries like South Korea and Canada produce a large amount of U.S. animated content. It will be done.

As a result of AI, physical and intellectual production tasks are likely to become increasingly automated, but the prospect of fully automated enterprises is also conceivable, where administrative functions such as pricing and acquisitions are also automated.

“It really depends on the type of company we’re talking about and all the operations involved, but certainly some of it is automated,” Frey said.