What is the Second Machine Age?

What Is the Second Machine Age?

The second machine age is the name Erik Brynjolfsson and Andrew McAfee gave to our current era in their 2014 book The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies. Their argument is simple and powerful: the first machine age, powered by steam and electricity, mechanized physical work — muscles were replaced by engines. The second machine age, powered by digital technology and artificial intelligence, is doing the same for mental work. Brains are being supplemented, and increasingly replaced, by computation.

The authors are careful to say what the phrase does not mean. It is not the end of work, nor the arrival of a robot utopia. It is an era of transformation in which the productive capacity of the economy grows enormously, but in which the benefits can accrue to a small fraction of people and firms. Their central question is whether this era produces broadly shared prosperity or a society divided between a small number of super-productive "superstars" and everyone else.

Why "second machine age" rather than "digital revolution"? Because they want to capture the scale of the change. The first machine age changed the trajectory of human civilization over the course of a century. They argue that digital technologies — exponential, cheap, and combinable — are doing something of comparable magnitude, but faster, and the second machine age is the frame for understanding what that means for work, wealth, and society.

Historical Background

The first machine age began in the late eighteenth century with the steam engine and culminated in the factory system, mass production, and the modern industrial economy. It took generations for the gains to spread, and the process was wrenching — but eventually it produced the great expansion of the middle class, public education, and modern welfare states.

The intellectual roots of the second machine age lie in the computing revolution. Gordon Moore's observation that computing power doubles roughly every two years became the exponential curve that underlies everything digital. The internet made information cheap and universal. Machine learning turned data into prediction. By the 2010s, digital technologies had moved from automating routine tasks to performing tasks that had seemed uniquely human — recognizing speech, translating languages, beating champions at games, driving cars.

Brynjolfsson and McAfee wrote The Second Machine Age at the peak of the early deep learning wave, following their earlier book Race Against the Machine. They updated the argument as generative AI arrived: what looked like the frontier of the second machine age in 2014 — language, reasoning, creativity — was precisely what the next generation of systems began to automate. The second machine age turned out to be even broader than its architects predicted.

Key Concepts

The three properties of digital technology are the engine of the era. Digital goods are exponential (they improve at compound rates), digital (they can be copied at near-zero cost), and combinable (they can be recombined into new innovations). These properties explain why progress feels so fast and why it defies the normal economic rules of scarcity.

Bounty and spread are the authors' framework for the two sides of the transformation. Bounty is the increase in goods, services, and possibilities — cheaper information, better medicine, new forms of creativity. Spread is the uneven distribution of that bounty — the growing gap between those who benefit from intelligent machines and those who are left behind. The two are not contradictory; both can grow at once.

Winner-take-all markets are a mechanism of spread. In digital markets, the best product can serve everyone at near-zero marginal cost, so a handful of firms — and a handful of workers with the right skills — capture outsized rewards while the rest see stagnant wages. The authors argue this dynamic is structural, not accidental, and requires a policy response.

The "race against the machine" is the employment question. The authors were early to warn that this time, automation targets cognitive work directly. Their nuanced claim is not that the machines win the race by definition, but that the race is real: workers whose skills complement the machines do brilliantly; workers whose skills compete with the machines struggle.

The productivity paradox is the puzzle that the second machine age's bounty has not fully shown up in official productivity statistics. The authors' explanation: new general-purpose technologies take time to diffuse, organizations must reorganize around them, and the payoff arrives in a lag. It is a warning against both complacency and despair when reading the economic data.

Contemporary Relevance

The second machine age is no longer a prediction; it is the environment we live in. Generative AI writes, designs, and codes; algorithms route logistics; prediction models guide medicine and finance. The authors' central question — whether bounty will be broadly shared — is being tested in real time as AI adoption races ahead of education, labor law, and social protection.

The inequality stakes are visible in the data. The top of the income distribution has captured a large share of recent growth, while middle-skill jobs have polarized into high- and low-skill work. The "superstar" dynamic is visible in finance, technology, and increasingly in creative fields. The debate over AI and jobs is, at bottom, a debate over whether the second machine age will be an era of shared progress or concentrated gains.

For policy, the second machine age implies a to-do list that the authors have been advocating since 2014: invest in education and reskilling, support entrepreneurship and competition, repair the safety net, and consider new forms of income support as automation spreads. Whether societies do these things is a political choice. The machines set the pace; the institutions set the destination.

Sources

• Brynjolfsson, Erik, Daniel Rock, and Chad Syverson. Artificial Intelligence and the Modern Productivity Paradox. NBER Working Paper 24001, 2017. https://www.nber.org/papers/w24001
• Acemoglu, Daron, and Pascual Restrepo. Robots and Jobs: Evidence from US Labor Markets. Journal of Political Economy, 2020. https://doi.org/10.1086/705716

Related Topics

• Will AI Replace Jobs?
• What Is the Future of Work?
• What Is Technological Unemployment?
• What Is Automation?
• The Second Machine Age
• Automation Topic