Why Isn’t New Technology Making Us More Productive?

"Innovations like cloud computing and artificial intelligence are hailed as engines of a coming productivity revival. But a broad payoff across the economy has been elusive.

For years, it has been an article of faith in corporate America that cloud computing and artificial intelligence will fuel a surge in wealth-generating productivity. That belief has inspired a flood of venture funding and company spending. And the payoff, proponents insist, will not be confined to a small group of tech giants but will spread across the economy.

It hasn’t happened yet.

Productivity, which is defined as the value of goods and services produced per hour of work, fell sharply in the first quarter this year, the government reported this month. The quarterly numbers are often volatile, but the report seemed to dash earlier hopes that a productivity revival was finally underway, helped by accelerated investment in digital technologies during the pandemic.

The growth in productivity since the pandemic hit now stands at about 1 percent annually, in line with the meager rate since 2010 — and far below the last stretch of robust improvement, from 1996 to 2004, when productivity grew more than 3 percent a year.

Economies grow not only by adding more capital and labor. Another vital ingredient is a nation’s skill in creating and commercializing innovation, which makes investment and workers more productive.

Seemingly small percentage gains in productivity can make a big difference in a country’s wealth and living standards over time. Even an additional 1 percent annual increase in productivity over a few years, to 2024, would generate an extra $3,500 in per-capita income for Americans, McKinsey & Company estimated in a report last year. The 3.8 percent average annual gain from 1948 to 1972 was the engine of the nation’s postwar prosperity.

Productivity is not a cure-all for economic ills. “Even if the optimism about this wave of digital technology proves justified, that does not mean there will be a real sharing of the benefits,” said Laura Tyson, a professor at the Haas School of Business at the University of California, Berkeley, and a chair of the Council of Economic Advisers in the Clinton administration.

But a less productive economy is a smaller one with fewer resources to deal with social challenges like inequality.

The current productivity puzzle is the subject of spirited debate among economists. Robert J. Gordon, an economist at Northwestern University, is the leading skeptic. Today’s artificial intelligence, he says, is mainly a technology of pattern recognition, poring through vast troves of words, images and numbers. Its feats, according to Mr. Gordon, are “impressive but not transformational” in the way that electricity and the internal combustion engine were.

Erik Brynjolfsson, director of Stanford University’s Digital Economy Lab, is the leader of the optimists’ camp. He confesses to being somewhat disappointed that the productivity pickup is not yet evident, but is convinced it is only a matter of time.

“Real change is happening — a tidal wave of transformation is underway,” Mr. Brynjolfsson said. “We’re seeing more and more facts on the ground.”

It will probably be years before there is a definitive answer to the productivity debate. Mr. Brynjolfsson and Mr. Gordon made a “long bet” last year, with the winner determined at the end of 2029. But studies at the industry and company levels, tapping data that ranges from Census Bureau business surveys to online job listings, show the pattern of technology diffusion and the obstacles.

The leaders are mainly large companies that have been investing in digital technology for years and high-growth younger companies, which are often backed by venture capital. Cloud computing is fairly widely adopted, but not the most advanced technology, like A.I applications.

The limited uptake, some experts say, is not so surprising at this stage, given that three-quarters of American businesses are small, with fewer than 10 employees.

At Anthem, a health insurer whose plans cover more than 45 million people, about 75 percent of the customer questions are now handled through its digital channels, including a web portal, a mobile app and speech recognition software. Three years earlier, the digital share was about 30 percent. The question-answering technology to help people with basic tasks like checking the status of a claim, paying a bill or finding a doctor is animated partly by A.I.

Digital automation has eliminated 10 million phone calls that Anthem’s call centers would have fielded, estimated Rajeev Ronanki, president of digital platforms.

Anthem, which is changing its corporate name next month to Elevance Health, is not cutting its customer service staff. But the role of those workers and how their performance is measured have changed. The traditional metric of performance in call centers is “call-handle time,” and the less time per call, the better. Anthem now wants its customer service staff to resolve problems for callers with one call, whenever possible, rather than passing them to another department.

Many of its call center agents have received additional training to become what Anthem calls “care navigators.” Measurements of their performance now include issues resolved and consumer satisfaction surveys. By that broader set of measures, Mr. Ronanki said, the company’s contact agents are 30 percent to 40 percent more productive. Adding skills and redesigning work, he said, are as important as improving technology.

“Building the technical capability alone is just the beginning,” Mr. Ronanki said.

It takes time for new technologies to spread and for people to figure how to best use them. For example, the electric motor, which was introduced in the 1880s, did not generate discernible productivity gains until the 1920s, when the mass-production assembly line reorganized work around the technology.

The personal computer revolution took off in the 1980s. But it was not until the second half of the 1990s that economic productivity really surged, as those machines became cheaper, more powerful and connected to the internet.

The 1990s revival was helped by a leap in technology investment by companies and by venture capitalists, especially in internet and web start-ups. Similarly, in the past decade software spending in the United States has more than doubled to $385 billion, as companies invest to digitize their operations, the research firm IDC reported.

Venture investment in artificial intelligence start-ups worldwide increased more than 80 percent last year to $115 billion, according to PitchBook, which tracks financing.

Cresta is an A.I. start-up trying to make a dent in the modern productivity problem. In 2020, Cresta introduced its initial product: real-time recommendation and coaching software for call center agents. Its technology digests huge volumes of text and voice conversations to identify patterns of behavior, and answers to questions that solve customer problems or generate sales.

The goal is not to replace workers but to lift their performance, said Zayd Enam, the company’s co-founder and chief executive. Cresta’s offering, he said, is made possible by recent advances in the power and speed of A.I. software, which he described as “game changing.”

Cresta has 200 employees, has raised more than $150 million in venture funding and has several dozen corporate customers including Verizon, Cox Communications and Porsche.

CarMax, the nation’s largest used-car retailer, started trying out the Cresta software in December. The A.I. experiment followed years of investment to shift the company’s computer operations to run on more flexible, cloud-based systems, said Jim Lyski, executive vice president for strategy, marketing and products.

Customer inquiries to CarMax’s contact centers tend to be lengthy. Used cars span different years, models, features and driving histories, and financing plans for what is a major purchase vary. The range of questions is all but unlimited, Mr. Lyski said, so purely automated communication is not an option.

But a computing assistant that could help sort all the automotive complexity, offering real-time suggestions and information, was appealing. Cresta first trained on the CarMax contact center data, and the experiment began with its live chat agents, who have text conversations with customers.

The experience has been encouraging, Mr. Lyski said. There has been about a 10 percent improvement in response time, conversion to sales and reduced session time. And the system keeps learning and getting better. The company has begun a pilot project with agents who field voice calls, lifting the total number of agents using the A.I. technology to 200.

One concern, Mr. Lyski said, was how employees would respond to having A.I. over their shoulders. Would it be good enough to be seen as a welcome helper instead of an irritating distraction? The reaction has been positive, he said.

Cresta began with contact centers as a large, early market because it is a labor-intensive field where A.I. can be applied relatively quickly and productively. But Mr. Enam sees its “real-time intelligence A.I.” potentially being useful in a wide range of knowledge work, acting as a clever assistant in everything from hiring to product development.

“This technology is more general purpose than we see now,” he said.

Mr. Brynjolfsson of Stanford is betting that’s true, and Mr. Gordon of Northwestern is doubtful.”