← Wiki

The Two Machine Ages

Overview

Brynjolfsson & McAfee’s framing of the two great technological inflection points in human history:

First Machine AgeSecond Machine Age
EraLate 18th century onwardLate 20th century onward
Key technologySteam engine (Watt, 1765–1776)Computer / AI
What it augmentsPhysical power (muscle)Mental power (cognition)
Scale”Made mockery of all earlier history” (Morris)Potentially comparable or greater
Curve-bendingPopulation and social development explodeDigital capabilities explode exponentially

Perez places these two ages within a broader five-revolution framework: the Industrial Revolution (1771), Steam & Railways (1829), Steel & Electricity (1875), Oil & Mass Production (1908), and Information & Telecommunications (1971). AI may represent the beginning of a sixth. See Technological Revolution Cycles.

The “Second Half of the Chessboard”

Moore’s Law means computing power doubles regularly. Like the legend of rice grains on a chessboard, the first half of the doublings produces modest results. We are now entering the second half, where each doubling produces staggering leaps. This is why AI capabilities seemed to arrive “suddenly” — exponential growth is deceptive until it isn’t.

The Repeating Pattern

Both Brynjolfsson/McAfee and Frey emphasize: the Industrial Revolution’s long-term benefits are uncontested, but the short-to-medium-term pain was severe. AI is following the same trajectory:

  • Middle-income job displacement
  • Labor share of income falling
  • Profits and capital returns surging
  • Growing political resistance

The difference: the first machine age primarily affected physical labor; the second affects cognitive labor — which means virtually every sector and skill level is potentially exposed.

Gordon’s Challenge: Breadth vs Depth

Robert Gordon argues the “special century” (1870-1970) was unique because it transformed everything simultaneously — food, housing, transportation, sanitation, communication, work. Post-1970 digital innovations changed only computing, communications, and entertainment. If AI similarly transforms only one dimension of economic life (cognitive work), it may produce less impact than electricity and automobiles combined. Counter-argument: cognitive work constitutes a larger share of the economy now than physical labor did in 1870, and AI agents are already expanding into physical-world coordination.

The Organizational Lag

Every machine age shares a structural feature: technology arrives decades before organizational redesign catches up. Gordon documents that electric motors were available for 40 years before factories figured out unit drive. David’s “dynamo and computer” parallel shows the same pattern for IT. For AI, the models may already be powerful enough; what’s missing is the organizational architecture to exploit them. See The Productivity Paradox.

Connection to Judgement

If the second machine age automates cognitive routine, what remains for humans is judgement — non-routine, context-dependent thinking AI cannot yet replicate. Acemoglu’s framework adds a political dimension: whether AI augments judgement or replaces workers depends on the direction technology is pushed, a social choice rather than a technological inevitability. See Judgement vs Knowledge in the AI Era and The Direction of Technology.