How Will Artificial Intelligence Affect Jobs?

TL;DR

Artificial intelligence will replace some tasks and jobs, but its broader employment effect depends on whether new demand and occupations offset that displacement. Tyler Cowen expects a slow, difficult adjustment marked by stagnant wages and concentrated gains, while Alex Tabarrok argues that cheaper technology raises wealth, spending, productivity, and demand for new kinds of human work.

Transcript

♪ [music] ♪ - [Alex] Tyler, a lot of people seem to be worried that the robots are coming: Siri, Cortana, artificial intelligence, Watson. They're going to take our jobs. So, do you think robots, are they going to take our jobs, and if they did, would that be a bad thing? - [Tyler] I've written on this topic and let me tell you, I'm worried for at ... Read More

Key Insights

• Modern automation is increasingly competing with human intelligence, not merely physical labor. Machines can perform tasks associated with medical diagnosis, legal research, journalism, games, and teaching, which gives Tyler Cowen reason to believe the current technological transition differs from earlier mechanization.
• Labor force participation is disputed as evidence of technological displacement. Cowen points to decades of decline and reduced participation among prime-age men, women after an earlier increase, and elderly workers, while Tabarrok attributes important parts of the historical pattern to broader societal change.
• Automation can create an income effect that offsets its substitution of machines for workers. Tabarrok argues that lower production costs make society richer, encourage additional spending, expand businesses, and generate new jobs that could not have been anticipated before the enabling technology existed.
• The ATM example shows how task automation can initially coexist with employment growth. ATMs replaced some teller work at individual banks, but cheaper branch operations supported more branches and additional tellers, although Cowen notes that teller employment later declined and branches began closing.
• Technology gains can be widely consumed while income gains remain concentrated. Tabarrok emphasizes that billions of people obtained powerful smartphones cheaply, while Cowen responds that measured real wages remained barely higher over decades despite including the value created by smartphones.
• Income distribution is central to the employment debate. Cowen argues that smart labor, smart capital, and intellectual property owners capture disproportionate gains, while passive labor and capital fare worse, leaving many middle-class Americans without meaningful real wage growth.
• Historical economic transitions are evidence for both optimism and caution. Tabarrok sees shifts from agriculture to manufacturing and then services as routes to better jobs, while Cowen emphasizes that adjustment took so long that agriculture continued receiving support long after industrialization began.
• Education and retraining are necessary but contested responses to automation. Tabarrok argues that online education and future enhancements can improve workers' capabilities, while Cowen points to flat graduation rates and test scores as evidence that many people struggle to work effectively with rapidly changing machines.

Questions & Answers

Q: How will artificial intelligence affect employment?

Artificial intelligence will replace certain human tasks and may eliminate some jobs, especially as machines move beyond physical labor into diagnosis, research, journalism, law, medicine, and teaching. Its total effect remains disputed because cheaper production can increase wealth, spending, business expansion, and demand for new occupations. The central question is whether those new opportunities emerge quickly enough and remain accessible to displaced workers.

Q: Why does Tyler Cowen think modern automation is different?

Tyler Cowen argues that machines now encroach on human intelligence rather than only replacing brute physical labor. He also points to declining labor force participation, stagnant real wages, a lower income share for labor, and slow worker reallocation supported by welfare and transfer payments. He expects a gradual loss of meaningful jobs and a difficult adjustment, not an apocalyptic takeover by robots.

Q: Why is Alex Tabarrok optimistic about automation and jobs?

Alex Tabarrok argues that past fears of technological unemployment repeatedly failed to account for the new demand created by greater productivity and wealth. Machines can substitute for workers in particular tasks, but lower costs allow businesses to expand and consumers to spend more. That process can produce new occupations, including jobs that were impossible to imagine before technologies such as smartphones became widespread.

Q: What does the ATM example show about automation?

The ATM example illustrates the difference between replacing a task and eliminating employment across an industry. An ATM substituted for teller labor within an individual branch, but it also made branches cheaper to operate. Banks could therefore open more branches and employ additional tellers. Cowen adds an important qualification: teller employment eventually declined, branches closed, and machines increasingly replaced in-person banking services.

Q: Does declining labor force participation prove robots are taking jobs?

Declining labor force participation is presented as suggestive evidence, not an uncontested proof. Cowen says participation has fallen for decades and highlights reduced work among prime-age men, women after their earlier increase, and elderly people receiving income support. Tabarrok responds that women's large entry into paid work and other social changes complicate the trend, making it difficult to attribute participation patterns primarily to robots.

Q: Who benefits financially from advances in automation?

The economists disagree about how broadly automation's gains are distributed. Cowen argues that smart labor, smart capital, and owners of intellectual property capture much of the value, while many middle-class workers experience stagnant real wages. Tabarrok emphasizes that consumers also benefit because powerful technologies become cheap and widely accessible, as shown by the spread of smartphones to billions of people around the world.

Q: Will automation create new types of jobs?

Automation can create jobs by reducing costs, increasing wealth, and encouraging additional spending. Tabarrok points to smartphone application development as work that did not exist before the underlying technology became available. He argues that future occupations will likewise be difficult to predict. Cowen accepts that some people will gain but doubts that most workers possess the education and adaptability required to capture those opportunities.

Q: How should workers and society respond to AI automation?

The proposed response is adaptation rather than destroying or rejecting machines. Cowen says society should recognize that the adjustment may be as traumatic and prolonged as the transition from agriculture to manufacturing. Both sides consider education and retraining important. Tabarrok believes technology, including online education, can strengthen those efforts, while Cowen warns that flat graduation rates and test scores reveal serious limits to present adaptation.

Summary & Key Takeaways

• Tyler Cowen argues that modern automation differs from earlier mechanization because machines increasingly perform cognitive tasks, including medical diagnosis, legal research, journalism, and teaching. He connects this change to declining labor participation, stagnant real wages, and a growing income share for capital, while predicting gradual rather than apocalyptic job losses.

• Alex Tabarrok argues that technological unemployment fears have repeatedly followed economic downturns without proving that machines permanently eliminate work. Automation can reduce employment in one task while lowering costs, increasing production, stimulating spending, and creating occupations that previously did not exist, as illustrated by bank branches and smartphone application development.

• Both economists agree that machines will profoundly change work and that education, retraining, and adaptation matter. Their disagreement concerns who will capture the gains and how quickly displaced workers can adjust. Cowen expects substantial hardship and inequality, while Tabarrok expects historical patterns of better, more productive, and higher-paying work to continue.