Exploring AI’s Economic Impact

Key Takeaways

Technology has always played a significant role in shaping the economy, with innovations driving economic growth and remaking industries. Each wave of tech innovation has left its mark.

As the world stands on the brink of an AI revolution, its potentially transformative impacts are generating immense interest.

How Has Technology Affected the Economy in the Past?

Annual U.S. growth in total factor productivity — a measure inclusive of innovation and technological change — averaged about 1.5% in the 1950s and 1960s. Except for a brief period after COVID-19 pandemic restrictions eased in 2021, it has since averaged about half that rate.

Massive government spending on highways, airports and other infrastructure helped fuel that mid-century boom. An expanding and increasingly educated workforce played a role, thanks to the GI Bill and evolution of the modern university system. Economists also credit changes in offices and production lines triggered by early automation and computers.

But corporate executives surveyed in 1957 about what accounted for boosts in labor productivity provided a resounding answer: air conditioning.¹ They credited an innovation not usually associated with sophisticated technological advancement.

In 1950, few homes had central or room air conditioning. By 1993, 68% had acquired it, and 90% by 2013.²

Consider that without air conditioning, the tremendous growth of Sun Belt cities, such as Atlanta, Dallas, Houston and Phoenix, might never have happened. Even mundane technological changes can contribute to economic growth in unforeseen ways.

Now imagine the potential impact of AI, a technology that many argue is set to revolutionize society. Economists, market analysts, scientists and consultants believe AI will fuel productivity gains not witnessed for generations. That accelerated growth would help alleviate inflation and potential labor shortages across the global economy.

How Are Labor Productivity and Inflation Related?

At a basic level, labor productivity measures a worker’s output relative to the labor necessary to generate it. Productivity increases when workers generate more output with the same effort. And when productivity increases, economic growth expands and living standards improve.

Next, consider the knock-on effects of rising productivity on inflation. When productivity increases, the amount of goods or services workers produce rises compared to the costs. As production rises, the overall supply of goods increases. At the same time, the costs of producing each additional product unit or service drop.³

More output and lower production costs place downward pressure on the prices of goods and services. When that happens — violà — inflation falls.

How AI Could Boost Labor Productivity

New research from ChatGPT creator Open AI, OpenResearch and the University of Pennsylvania concludes that AI will change most jobs somehow.⁴ Its potential to increase workers’ capacity strongly argues for rising productivity in coming years.

Goldman Sachs estimates that AI will boost annual U.S. labor productivity by 1.5 percentage points in the first decade after its widespread adoption.⁵ That would mark the most significant increase in a quarter century.

Meanwhile, 81% of U.S. businesses now employ the technology, up a third in just five years.⁶ This includes using AI-powered programs such as Grammarly, a tool designed to help ensure content readability. But such programs pale compared to the potential of ChatGPT, which can produce several types of original content instead of merely editing existing text.

A 30-fold increase in AI patent filings between 2015 and 2021 indicates businesses will encounter additional ways to employ AI in coming years.⁷ Of course, how they use and achieve AI’s productivity potential will vary from industry to industry and company to company.

Automating all or a portion of our daily routines will account for some gains. Consultant McKinsey & Co. estimates AI could automate a quarter of all work activities by 2030. In coming decades, 60% of the 800 occupations tracked by the U.S. Bureau of Labor Statistics could have at least a third of their activities automated.⁸

For example, Microsoft’s new CoPilot AI “promises to unlock productivity” by summarizing long email threads or drafting suggested email responses. It may also automate scheduling meetings and notetaking while enhancing other Office suite programs to make work more efficient. Similarly, Google plans AI enhancements to its Cloud and Workspace apps.

AI Could Displace Jobs, but New Roles Will Emerge

Naturally, the automation opportunities AI presents have spurred fears of widespread job displacement or elimination. This concern isn’t unfounded: According to one estimate, the retail sector could see a third of its jobs displaced by 2030.⁹

But history has shown that for all the jobs lost to technological advancement — steam engine mechanics, toll booth operators and switchboard operators, to name a few — others will emerge.

One study found that 60% of today’s workers occupy jobs that didn’t exist in 1940. Since then, 85% of employment growth has consisted of new jobs that technological innovation created.¹⁰

Boston Consulting Group and Faethm, a firm specializing in AI and analytics, project that one additional job will surface for every six jobs AI automate or augment. Their study predicts that new roles developing, implementing and running new AI programs will appear in nearly five dozen occupations.¹¹

For instance, companies have already begun hiring “prompt engineers” to communicate effectively with large-language AI systems.¹² ZipRecruiter reports that “hundreds” of additional job postings mentioning AI appear monthly. Figure 1 looks at the demand for AI labor across industries.

Figure 1 | AI Labor Demand Is Increasing

^{Source: Lightcast, 2022 | Chart: 2023 AI Index Report, Stanford University.}

Workers in existing roles can undertake higher-value tasks while AI manages more tedious, ordinary duties. These include certain clerical, administrative, computational and manual duties in various blue- and white-collar positions.

Aiding the transition — robocolleagues that consulting firm Gartner Inc. expects to interact with 100 million workers by 2026.¹³

AI Could Help Address Global Labor Shortages

Far from causing widespread job losses, AI may help fill the gap from approaching global labor shortages created by demographic shifts and other variables. The aging of the Baby Boom generation and uncertainty about future immigration policies present economic challenges. The U.S., Japan and many other developed economies confront ever-shrinking working-age populations.

This slowing, or negative, population growth resulting in fewer workers relative to the non-working-age population is known as a “demographic burden.” But AI offers societies and economies another way to help shoulder this burden. Specifically:

• In the U.S., the Boston Consulting Group (BCG) and Faethm foresee a workforce shortfall of 0.9-4.4% by 2030. Widespread AI adoption between now and then could close that gap. Low AI adoption, on the other hand, would leave a deficit of 12.5 million jobs. See Figure 2.

• The BCG/Faethem study also forecasted occupations with the highest number of job surpluses and shortfalls due to AI adoption. For example, researchers believe AI will create a surplus of information and record clerks. At the same time, they project a shortfall of qualified workers to fill computer occupations.

• In a best-case scenario, Goldman Sachs predicts AI-facilitated robots could fill at least 48% of a projected U.S. manufacturing labor shortage by 2030. Similarly, AI-enabled productivity gains could narrow an expected gap in care for older people by as much as 53% by 2035.¹⁴

Figure 2 | Projected U.S. Worker Surpluses and Shortfalls in 2030

^{Source: Faethm, BCG analysis.}

AI May Boost Global GDP by Trillions

The productivity improvement AI drives should directly boost economic growth.

During the next decade, Goldman expects AI to increase global gross domestic product (GDP) by 7%, equaling about $7 trillion. Bank of America is more optimistic. It estimates a potential $15.7 trillion benefit.¹⁵

AI could expand the economy profoundly in the U.S, doubling annual GDP growth potential to 3%.

This elevated growth could include varied impacts:

• A global AI market — software, hardware and services — reaching $900 billion by 2026 and $2 trillion by 2030.¹⁶

• Soaring private investment, which doubled to $94 billion in 2021 from just a year earlier.

• Health care spending savings of up to $360 billion annually in the next five years.¹⁷

• Consumer spending via chatbots reaching $142 billion as early as 2024.¹⁸

• Declining costs for AI technology and training:

  • Median prices for robotic arms have declined 46% in the past five years.

  • Training costs for image classification systems have dropped 64% in the past five years, while training times have declined by 94%.

AI’s Productivity Jump Offers Hope in Post-Pandemic World

AI could boost productivity in a manner unmatched in recent memory. Its accruing economic benefits may relieve a world whipsawed by the worst pandemic in a century and surging inflation afterward. Longer term, it could help developed economies avoid the worst economic effects of aging populations.

More than that, AI’s inevitable assimilation into daily life might nudge society nearer to the prophecy that legendary economist John Maynard Keynes outlined in his famous 1930 essay, “Economic Possibilities for Our Grandchildren.”

Keynes envisioned a world in which technology eventually rid humans of unnecessary work, setting them free for more noble pursuits:

Keynes, of course, wrote these words in an office with no air conditioning, but his intelligence was anything but artificial.