AI chatbots are among the tools that could jump-start US productivity. |
Artificial intelligence provides a potential significant boost to US productivity at a time when demographic and labor supply challenges are restraining economic growth.
|
Productivity growth slowed after the Great Recession, limiting economic growth across the developed world. After a brief spur at the start of the COVID-19 pandemic, US labor productivity — measured by worker output per hour — has returned to its trend growth rate of about 1.25%, according to economists at S&P Global Market Intelligence. That is below the pre-pandemic level of above 2%, and the gap amounts to trillions of dollars in lost economic output by some estimates.
Companies including Alphabet Inc. and Microsoft Corp. are spearheading the use of generative AI in a variety of ways, primarily through the creation of humanlike text and realistic images, while others are broadly exploring the potential of the technology as a timesaving and efficiency tool. Further advances in AI could bolster US productivity by 1.5 percentage points in the 2030s, according to research by consultancy Capital Economics.
"We're seeing that it does improve productivity among workers," said Yong Suk Lee, assistant professor of technology, economy and global affairs at the University of Notre Dame, adding that workers are seeing the time they need to finish their work decline and the quality of it increase with exposure to ChatGPT, the AI-powered language model developed by OpenAI LLC.
Productivity on the go-slow?
The recent lull in productivity growth contrasts a longer-run average of 2.2% between 1948 and 2019, according to consultancy McKinsey & Co. Output has largely stayed below 1.5% since 2005 following a surge of 3.3% annually on average between 1996 and 2004, fueled by advances in information technology.
"If the US economy were a car, the engine has been sputtering for a while," McKinsey consultants wrote in a February report.
Since the Great Recession, growth in real value-added output by employees has been driven by increases in the number of hours worked, rather than increased output per hour. Labor productivity picked up a little after 2015 and improved further during COVID-19 as millions of jobs were lost in low-productivity sectors such as retail and restaurants. But as those jobs returned, this improvement in productivity reversed.
"The recent plunge in productivity has essentially reversed the pandemic surge, leaving the recent level in line with the rising trend that was in place starting around 2015," said Joel Prakken, chief US economist at S&P Global Market Intelligence.
The need to boost productivity has been a major concern for policymakers around the world and contributed to the massive package of tax breaks and subsidies under President Joe Biden designed to boost investment and output. McKinsey anticipates a return to an annual output growth of 2.2% would be worth $10 trillion to the US economy across a decade.
The real risk to the economy is that productivity growth does not increase to make up for the constraints on the labor supply. The US birth rate has been below the replacement rate of 2.1 births per woman since 2007, reaching a more than 60-year-low of 1.64 in 2020, according to the World Bank. The prospect of fewer people entering the workforce in the future is compounded by greater political resistance to immigration, as well as a long-term decline in the participation rate of working age adults in the 21st century.
Should generative AI bolster worker output, it "would mean a gradual but striking end to the period of low productivity growth which has dogged developed economies for most of this century," Neil Shearing, group chief economist at Capital Economics, said in a Sept. 26 report.
But the technology is also being met with a fair amount of skepticism.
"I'm reminded of [American economist] Robert Solow's famous quip about the digital revolution: 'It's evident everywhere but in the productivity statistics,'" Market Intelligence's Prakken said. He is not basing his long-term productivity growth outlook on recent AI developments "because I have Solow's quip in mind."
AI uptake
Alphabet's Google Maps, Apple Inc.'s Siri voice assistant and Amazon.com Inc. online product recommendations are among many AI-powered applications used by households and businesses. Microsoft products based on OpenAI's ChatGPT are now used by millions.
"Most organizations that are experimenting with these tools — such as ChatGPT — are just using the public facing tech," said Alexander Johnston, a data, AI and analytics analyst at Market Intelligence. Johnston said 40% of the organizations Market Intelligence surveyed were using the technology to some degree but generally not in a "particularly sophisticated manner."
AI is applicable in many data-heavy functions or manufacturing industries that use a lot of sensors and tool recognition, Johnston said. The biggest impact is likely to be services such as fraud detection in finance.
"This revolution is already well underway in the software industry and, within the next few years, will spread quickly in other areas such as education, translation services and medical diagnostics," Shearing said.