In previous issues, we've looked at how artificial intelligence (AI) will transform such industries as:
- Transportation (with driverless cars and trucks)
- The military (with robotic soldiers and sentries)
- Healthcare (with tools like wireless health monitoring and IBM's Watson for Oncology, which helps doctors improve the accuracy of their diagnoses)
- Advertising (with algorithms that determine which ad to show to a shopper based on her past purchases, preferences, and demographics)
- Financial services (with knowledge engines and digital assistants that answer customers' questions in plain English)
- Manufacturing (with factory robots that work side by side with human workers)
- Law enforcement (with facial recognition software that identifies criminals from surveillance footage)
Now, it's time to examine the big picture by putting all the pieces together to see how artificial intelligence will affect the economy. According to The McKinsey Global Institute, AI's transformation of the economy is "happening ten times faster and at 300 times the scale-or roughly 3,000 times the impact" of the Industrial Revolution. As AI transforms one industry after another over the coming decade, it will spur economic growth throughout the world, but particularly in the United States.
This couldn't happen at a better time. According to calculations based on World Bank data, in the six full years since the Great Recession ended in mid-2009, U.S. GDP has grown by a woeful average of only 2.1 percent per year.
Now, as AI enables businesses and individuals to increase their productivity, the U.S. will finally have the means to achieve average GDP growth of 3.5 percent or greater that we've been predicting, starting as soon as 2017 and lasting for at least the next twenty years.
Even that prediction may be too conservative, according to a new study released by the Accenture Institute for High Performance, in collaboration with Frontier Economics, titled "Why Artificial Intelligence is the Future of Growth."
The study looked at the potential impact of AI on economic growth, focusing on the twelve countries that account for more than half of the world's economic output. It found that AI technologies will increase labor productivity by as much as 40 percent by changing how work is performed and creating a new collaboration between people and machines in the workplace.
According to Paul Daugherty, Accenture's chief technology officer, "Our research demonstrates that as AI matures, it can propel economic growth and potentially serve as a powerful remedy for stagnant productivity and labor shortages of recent decades."
Specifically, the study found that AI could increase productivity by an average of 40 percent, while doubling annual economic growth rates for most countries by 2035.
The researchers contrasted a baseline scenario for each country's economy in 2035, which depicts projected economic growth without AI, versus an AI scenario, in which AI has been absorbed into the economy. They found that AI made the biggest impact on the U.S., propelling its annual growth rate from 2.6 percent to 4.6 percent by 2035, which the country hasn't experienced since the boom years of the 1980s. That's equivalent to an extra $8.3 trillion in gross value added (GVA) which is a close approximation of GDP. That increase would amount to the current GVA of Germany, Japan, and Sweden combined.
This estimate is in line with a study recently released by Analysis Group, Inc. With funding from Facebook, the researchers studied the potential impact of AI on the world economy and concluded that it will add from $1.49 trillion to $2.95 trillion to global GDP over the next ten years.
According to Accenture, other countries will also benefit from deploying AI. While none of them are projected to achieve the 4.6 percent annual growth rate of the U.S. economy, the size of their increase is in many cases greater because their baselines are lower. Many countries-such as Austria, Finland, Germany, the Netherlands, and Sweden-will double their economic growth, while Japan's gain will be more than triple, from 0.8 to 2.7 percent.
However, improvements in productivity and innovations won't help the quality of life if they come at a cost of massive job losses. Many observers have predicted that AI software, driverless vehicles, and smart robots will make entire occupations obsolete and put millions of people out of work. For example, Bank of America Merrill Lynch recently predicted that by 2025 businesses will save $9 trillion in employment costs as AI automates knowledge work, and manufacturers and healthcare companies will save $8 trillion as AI takes over human jobs.
As pointed out in The Wall Street Journal's "CIO Journal" blog, this is the same argument that is raised every time a major new technology emerges. In 1821, during the Industrial Revolution, British economist David Ricardo wrote that "the employment of machinery is frequently detrimental to [human workers'] interests." A century later, in 1930, British economist John Maynard Keynes diagnosed what he called "a new disease" that he termed "technological unemployment," defined as "unemployment due to our discovery of means of economizing the use of labor outrunning the pace at which we can find new uses for labor."
But despite these fears, according to The Economist, in every previous technological revolution, more new jobs were created than were destroyed. MIT economics professor David H. Autor reaches the same conclusion in his research paper published in the Journal of Economic Perspectives. He points out that in 1900, 41 percent of U.S. workers were employed on farms. As tractors and other agricultural machinery replaced human workers, the industry employed only 2 percent of the American workforce by 2000. Similarly, automation has replaced many human factory workers, and computers and software eliminated countless office jobs.
Yet despite all these disruptions to the labor force, as of November 2016, according to the latest data from the Bureau of Labor Statistics, 159,486,000 Americans currently hold jobs. How can new technologies constantly destroy millions of jobs, while the vast majority of workers are still able to find work?
According to Autor, the explanation is simple: "Tasks that cannot be substituted by automation are generally complemented by it. Most work processes draw upon a multifaceted set of inputs: labor and capital; brains and brawn; creativity and rote repetition; technical mastery and intuitive judgment; perspiration and inspiration; adherence to rules and judicious application of discretion. Typically, these inputs each play essential roles; that is, improvements in one do not obviate the need for the other. If so, productivity improvements in one set of tasks almost necessarily increase the economic value of the remaining tasks."
As the more routine parts of jobs become automated by technology, workers can spend more time on the other parts of those jobs that add more value to customers. As Boston University School of Law economist James Bessen recently discussed in The Atlantic, "It turns out that workers will have greater employment opportunities if their occupation undergoes some degree of computer automation. As long as they can learn to use the new tools, automation will be their friend."
To illustrate this concept, consider that automated teller machines (ATMs) were widely expected to eliminate the jobs of most bank tellers when they began to be installed in the 1970s. From 1995 to 2010, the number of ATMs in the U.S. soared from 100,000 to 400,000, which would lead to the assumption that jobs for human tellers declined just as rapidly. But that's not what happened. Between 1980 and 2010, the number of tellers actually increased by 10 percent, from 500,000 to 550,000.
Why did this occur? According to Bessen, the use of ATMs lowered the cost of operating a bank branch, so the number of bank branches increased by more than 40 percent from 1988 to 2004. While the number of tellers per branch dropped by more than 33 percent, the larger number of branches meant that more tellers were needed overall.
At the same time, as ATMs took over many of the routine transactions that tellers once handled, the use of information technology allowed them to offer "relationship banking" services, by getting to know customers on a personal level and telling them about other bank offerings such as credit cards, CDs, mortgages, automobile and student loans, and investment vehicles.
Bessen found the same pattern playing out in one occupation after another, from supermarket checkers to graphic designers. Instead of destroying jobs, computers and automation changes them in ways that lower costs and increase demand. He concluded that "Rather than contributing to unemployment, the number of workers in these occupations grew... On average, since 1980, occupations with above-average computer use have grown substantially faster (0.9 percent per year)."
The same pattern is likely to play out as AI takes over parts of jobs that are better suited to smart machines and software. For example, we predict that if artificial intelligence eliminates the time that physicians need to read medical journals and increases the accuracy of doctors' diagnoses, they'll have more time to spend talking to patients.
Patients who have better experiences visiting doctors and better results from more precisely targeted treatments will be more likely to visit doctors more often, and patients who currently avoid going to doctors will be more likely to overcome their reluctance to make an appointment. Demand for doctor appointments will go up, but doctors will become more productive as AI handles everything from scheduling appointments in order to reduce waiting times, to voice-recognition apps that will eliminate the time-consuming, error-prone process of handwriting medical records and prescriptions.
Examples can be found in nearly every industry that will be affected by AI. For example, in the world of luxury retail, a Technology Vision 2016 report cites the case of Moda Operandi. The store, whose customers spend $1,200 per order seven to eight times per year according to a recent New York Times article, delivers customized recommendations in a private shopping experience. When Moda Operandi created a new personalization engine, it enabled each stylist to serve three hundred customers, compared to as few as fifty previously.
Now it can provide the same high-end service to all of its customers instead of just the ones who spend the most. Naturally, this elevated level of service to less frequent buyers encourages them to spend more per visit, and to visit more frequently-and because of that increased demand, the stylists' jobs are in no danger of being eliminated.
Tying this idea back to our discussion on economic growth, we can see that artificial intelligence will deliver benefits such as increased productivity, while avoiding the threat of high unemployment that could create a drag on consumer spending that would otherwise offset those benefits.
Looking forward, we foresee the following developments emerging from this crucial trend:
First, artificial intelligence will lead to faster economic growth through intelligent automation.
Consider Amelia, an artificial intelligence agent developed by IPsoft. According to Accenture, it absorbs all of the maintenance manuals for an industry's equipment. When an engineer in a remote location describes a problem with a machine, Amelia uses natural language processing to understand what the engineer is saying, and then it figures out how to solve the problem and provides step-by-step instructions for fixing it. But Amelia isn't limited to a single industry.
For example, it has learned the correct answers to the 120 most commonly asked questions by mortgage brokers, and a bank has used it to save its loan officers' time by outsourcing the answering of those questions to Amelia. Amelia also recognizes situations in which it can't provide an answer; after forwarding the question to a human, Amelia pays attention to the answer and uses this knowledge the next time the question is asked. This self-learning capability of AI means that it is always improving, as opposed to other technologies that depreciate over time.
Second, AI will increase the productivity and effectiveness of workers across industries via labor augmentation.
AI will make workers more capable by equipping them with knowledge they could not acquire on their own. Consider Praedicat's AI platform, which leverages machine learning and big data. This helps personnel at insurance firms make better decisions. It digests all the information from 22 million peer-reviewed scientific research studies to predict which risks will pose the biggest threats to insurers. This allows underwriters to create policies that are based on an accurate assessment of risk, and to develop new insurance products to cover customers in situations that competing insurance firms haven't even considered.
Third, as AI transforms industries, those innovations will continue to ripple throughout the economy, creating further innovations.
For example, driverless cars won't just transform the automotive industry. Accenture theorizes that mobile service firms will enjoy increases in demand as riders in autonomous vehicles, freed from the responsibilities of driving, will have more time to use the Internet. Cities could use the data from driverless cars to charge riders on a pay-per-use basis, increasing prices on busy roads to discourage congestion and motivate the use of alternate routes.
Fourth, to prevent job losses, employers and educators will work together to close the "skills gap."
As long as employees have the right skills to work with AI, their roles will expand and their jobs will become more meaningful as they focus on activities that add value, while offloading routine tasks to machines and software. According to The Economist, "AI will not so much replace workers directly as require them to gain new skills to complement it." Workers will be given opportunities to master new skills through online learning, courses at community colleges and business schools, and on-the-job training.
Fifth, the deployment of AI in the manufacturing industry will accelerate the trend toward reshoring.
As productivity rises and costs plunge, American companies will be able to afford to move their production back to the U.S., which will help to offset any job losses in other sectors. However, according to a report from the Pardee Center for International Futures prepared for the U.S. National Intelligence Council, this means that developing countries will no longer be able to follow the path to modernization of their economies by developing cheap manufacturing capabilities for export markets.
It now costs the same amount to deploy a manufacturing robot as it does to outsource a manufacturing job to the average Chinese laborer. Boston Consulting Group found that more than one-third of companies with revenues above $1 billion are considering reshoring. As a result, developing countries will need to find another way to boost economic growth, and according to the Pardee Center report, there is no clear roadmap for them to follow.