Andrew McAfee

In the [YOU] on AI Field Guide

The cycle that began with [YOU] on AI asks what it would mean to see the machine clearly—without the narcotic of hype or the paralysis of fear. McAfee is the cycle’s most disciplined empiricist: he arrives at large conclusions only by way of small, verifiable facts, accumulated patiently until they compel a verdict. Edo Segal’s question—whether we are building the world the power makes possible or merely the world it makes profitable—is McAfee’s question stated in a different register. McAfee gives it an empirical edge: the bounty will come regardless of what we choose. Whether it reaches the many or pools among the few is a question the technology leaves open, and that we must answer.

His framework of bounty and spread is the cycle’s clearest account of why the same forces that generate unprecedented abundance also generate unprecedented inequality. The bounty side of modern AI is visible everywhere: cheaper expertise, cheaper analysis, cheaper creative work, the very things that were once scarce and expensive precisely because they required trained human minds. The spread side is the mechanism by which that bounty concentrates: the capital-labor split deepens as the value that once flowed to millions of cognitive workers flows instead to the much smaller number who own and deploy the systems that now perform that work.

The virtuous cycle he identified—better product attracts more users, more users generate more data, more data produces a better product—is the precise mechanism by which the data network effect concentrates AI returns. McAfee described it as a strategist helping his clients win, but the description doubles as a diagnosis of the structural inequality the technology tends to produce. His four horsemen of the optimist—capitalism, technological progress, public awareness, and responsive government—provide the cycle’s most systematic account of what it would take to capture AI’s bounty without letting its spread run unchecked.

McAfee’s deepest contribution to the cycle is his conviction that the trajectory of technology is not fixed by the technology. He has shown, across centuries of data, that the same technology has produced wildly different outcomes depending on the institutions that governed its deployment. The first machine age was genuinely brutal, and yet it eventually produced broadly shared prosperity—not because the technology changed, but because societies built the institutions that redirected its gains. This conditional optimism, earned against the evidence rather than in spite of it, is what the cycle most needs from him: the demonstration that the hopeful answer is not guaranteed but remains possible, and that the act of pursuing it is worth undertaking.

Origin

McAfee came to the study of technology through the unglamorous discipline of management research, asking how digital tools actually change what firms do and what workers earn. His training was in operations and information systems; his doctorate from Harvard Business School; his home for most of his career the MIT Sloan School of Management, where he co-founded and co-directs the Initiative on the Digital Economy. That institutional detail matters because it locates him precisely: he sits at the intersection where the abstractions of computer science meet the brute facts of payrolls, productivity statistics, and balance sheets.

The body of work he built with Brynjolfsson defines his intellectual profile. Race Against the Machine (2011) sounded an early alarm that technology was beginning to destroy jobs faster than it created them. The Second Machine Age (2014) expanded that into a full account of the digital transformation and became one of the defining texts of the decade. Machine, Platform, Crowd (2017) turned from diagnosis to strategy, mapping how the balance of power was shifting inside organizations. More from Less (2019) made the surprising environmental case that growth and resource consumption had begun to part ways in wealthy economies. The Geek Way (2023) examined the cultural norms behind the most successful technology companies. Read together, these books trace a single mind working out the implications of one enormous fact: that intelligence itself has become something we can manufacture.

What separates McAfee from both enthusiasts and doomsayers is his refusal to treat the technology as an autonomous force with a will of its own. The machine does not decide who benefits from it. That decision is made by people, through markets and laws and norms and the thousand small choices that add up to an economy. The same technology can be deployed to augment workers or to replace them, to broaden prosperity or concentrate it. The outcome is not a property of the silicon. It is a property of the society that wields it.

Key Ideas

The Second Machine Age. McAfee and Brynjolfsson argue that computers and digital advances are doing for mental power what the steam engine and its descendants did for muscle power. Three forces drive the transformation: exponential improvement in computing power, which takes the technology into the second half of the chessboard where the numbers grow so large that intuition fails; digitization, which allows any information good to be copied perfectly at essentially zero cost; and combinatorial innovation, in which each new digital technology becomes a building block recombined with every other, so the number of possible new inventions grows not arithmetically but explosively. Together these forces explain why progress in the second machine age feels not merely fast but accelerating.

The Great Decoupling. For most of the postwar period in the United States, productivity, economic output, employment, and median family income rose together in near-perfect lockstep. Beginning in the late twentieth century, the lines came apart. Productivity continued its steady climb, but the income of the typical family stopped tracking the economy’s gains. McAfee named this divergence the great decoupling—the dark twin of his optimism about the second machine age. The same technological forces generating unprecedented bounty were simultaneously severing the historic link between that bounty and the well-being of the median worker. The decoupling is not a destiny encoded in the technology; it is the outcome of how the technology has been deployed within a particular set of economic arrangements, and those arrangements can change.

Bounty and spread. McAfee and Brynjolfsson use two words to capture the double-edged character of the second machine age. Bounty is the good news: the staggering increase in the volume, variety, and quality of what the economy produces, and the steady fall in its cost. Spread is the bad news: the growing differences in economic outcomes among people, the widening gaps in income, wealth, and opportunity. The same forces producing the bounty are widening the spread, and the two are linked, not independent. A society that captures the AI bounty and distributes it broadly will experience the technology as a blessing; one that lets the spread run unchecked will experience the same technology as a calamity.

Racing with the machine. McAfee’s prescription is to race with the machine rather than against it. To race against a system that improves exponentially is a contest with a foregone conclusion. To race with the machine is to find the complementary tasks—the things humans do that the machine does not—and to pair human and machine so that each amplifies the other. His favorite illustration is freestyle chess, where human-computer teams outperform both grandmasters and supercomputers alone. The skill that matters most is the skill of the partnership, and the question to ask is not whether the machine will replace you but which parts of your work it can absorb and which become more valuable once it does.

The four horsemen of the optimist. In More from Less, McAfee identifies four forces that together produce good outcomes from technological progress: capitalism and technological progress supply the engine of bounty, while public awareness and responsive government are the forces that check harms the market has no incentive to address. Applied to AI, the framework directs us to ask whether the trailing horsemen—public awareness of AI’s risks and governmental capacity to respond—are keeping pace with the engine of capitalism and technological progress that is racing ahead.