The Hundredth EyeFR
AI & Jobs · Essay 5 / 5

Where Are We on the Wave of Progress?

Progress rises in waves, and we are not all at the same point on the wave.

November 22, 20265 min readLire en français

Progress comes in waves, and averages hide the people living through the trough.

In brief
  • Life expectancy has more than doubled since 1900, but 1918, 1929 and 2020 were deep troughs.
  • ChatGPT, launched in late 2022, had 700 million weekly users by the summer of 2025, and I wonder whether society will have time to adapt.
  • Western Europe produces twelve times more per person than sub-Saharan Africa, and with AI, some people rise while others sink.
  • I think AI is only at the start of installation; what we don't know yet is who the trough will hit, and for how long.

In 1910, a book claimed that war no longer paid.

Its author, the British journalist Norman Angell, argued that Europe’s economies had become so intertwined that no country could get rich by conquering its neighbor. The Great Illusion sold more than two million copies between 1910 and 1913, in 25 languages.1 Angell never said war was impossible. He spelled it out: “it is not the likelihood of war which is the illusion, but its benefits.”2 In 1914, Europe went to war.

The war of 1914 doesn’t prove Angell was wrong about the economics. It proves that being bound together, with every reason not to fight, wasn’t enough. My last piece made the same point: dependence forces countries to talk, but it doesn’t stop a race. And that’s where an idea I’ve carried for a long time begins: progress doesn’t rise in a straight line.

A rising wave

I picture progress as a sine wave, the curve that rises and falls like a wave, laid on a slope. The slope points up: over the long run, humanity lives better. Average life expectancy worldwide went from 32 in 1900 to 73 in 2023.3 The number of people in extreme poverty fell from about 2.3 billion in 1990 to an estimated 830 million in 2025.4

?quality of life →time →
Quality of life as the UN's Human Development Index measures it: health, education, income. Source: Argus Kalman.

Around that slope, the curve swings, and the troughs are deep. The 1918 flu killed at least 50 million people.5 In the United States, output fell by about a quarter between 1929 and 1933.6 In 2020, the world economy shrank by 3%, and global life expectancy dropped by almost two years between 2019 and 2021.7 Seen from afar, the curve rises. Seen from the bottom of a trough, it collapses.

On the same curve, one has an incredible view, the other sees only a wall that seems impossible to climb.

Waves closer and closer together

The second part of my idea comes from observation more than measurement, because the historical data isn’t reliable enough: I think the waves are coming closer together. For technology itself, the signal is strong. ChatGPT, launched at the end of 2022, had 700 million weekly users by the summer of 2025.8 I can’t think of a tool this big that spread this fast.

For crises, the data is murkier. The World Bank counts five global recessions since 1975: 1975, 1982, 1991, 2009 and 2020. The gaps between them, 7, 9, 18 and then 11 years, show no steady speed-up.9 So I can’t prove the troughs are coming closer. That leaves the question that worries me: if technology changes faster and faster, will society have less time to adapt before the next wave?

Not all at the same point

We are not all at the same point on the curve at the same time. Between countries, the gap is still enormous: in 2022, the wealth produced per person was on average twelve times higher in Western Europe than in sub-Saharan Africa.10 Some countries are in a trough while others climb.

The same goes within a country, and it’s the thread running through this series. AI delivers on its promises, but not for everyone at once. The person whose job disappears is in a trough, while the company that replaced them is on the way up. Someone who builds a store in an evening with AI may climb, while the shop down the street may slide. The average rises. But nobody lives in that theoretical average.

Where are we?

The economist Carlota Perez has studied the great technological revolutions, from railways to computing. In her view, each one goes through two main phases, separated by a crisis. First comes installation: money floods in, bubbles inflate, the rules lag behind. Then the bubble bursts, opening a period of crisis she calls the “turning point”: the rules have to be rewritten, and how long the crisis lasts depends on how capable governments and society are of doing it. Finally comes deployment, when the technology benefits far more people. For Perez, computing is a single revolution in three stages: first chips and computers, from the 1970s; then the internet, which made globalization possible, from the mid-1990s; and now generative AI, since the 2020s. She believes the world economy is at that turning point today, in a moment much like the 1930s.11

I read it a little differently: for AI, I think we’re still only at the start of installation. Money is pouring in: the big tech companies have announced hundreds of billions of dollars of investment in data centers. Yet in 2025, in the sectors Eurostat covers, only one in five EU companies with at least ten employees used AI, up from one in seven a year earlier.12 As the first piece noted, Andrej Karpathy estimates that only about twenty million professionals use AI for complex tasks.13 That lag is no surprise. As the first piece noted with electricity, it took decades, and factories redesigned around it, before the gains showed up.14 If, in a few years, most companies use AI every day, that will be the sign that real adoption, the deployment phase, has begun. It still won’t tell us who gets the gains.

How long before the recovery?

The move from installation to deployment is often marked by a crisis. If a trough comes, how long will it last? The great troughs of the last century lasted years, not centuries. In the United States, after the 1929 crash, output got back to its pre-crisis level in 1936.6 After the pandemic, global life expectancy was back to its pre-pandemic level by 2022.15 But an average that bounces back doesn’t give back the lives lost, or the careers broken.

For those living through it, the trough can last far longer than the average. In Britain, output per worker rose 46% between 1780 and 1840. Real wages rose only 12%, according to the economic historian Robert Allen, and barely moved from 1800 to 1830.16 For a generation, in the middle of the industrial revolution, the average worker’s wage all but stalled. Historians still debate these figures, and other estimates are less pessimistic.16

Nor is there any guarantee the slope will keep rising. The economists Daron Acemoglu and Simon Johnson, 2024 Nobel laureates, show in Power and Progress that throughout history, the gains from a technology were widely shared only when rules and countervailing power forced it.17 Recovery is a collective effort. Time alone won’t bring it.

I think it will come. Throughout this series, the same question has kept coming back in different forms: who will buy, who will invent, who will lead the race. AI will most likely deliver on its promises. The question is who the trough will hit, and for how long.

To borrow Angell’s phrase, the great illusion of our time may be the belief that when the world grows richer, everyone benefits at the same time.

The curve of human progress will very likely keep rising. Unfortunately, an average is not a life, and some will live through the trough.

End of the series. The eye stays open, searching for the signal.

Sources

  1. Weinroth, “Norman Angell and The Great Illusion: An Episode in Pre-1914 Pacifism,” The Historical Journal, 1974 (over 2 million copies 1910–1913, 25 languages) — https://www.cambridge.org/core/journals/historical-journal/article/abs/norman-angell-and-the-great-illusion-an-episode-in-pre1914-pacifism/2F3411902E4AC3DC5BEE90BA8D6CACE9

  2. Angell, The Great Illusion, 1913 edition, Project Gutenberg — https://www.gutenberg.org/files/38535/38535-h/38535-h.htm

  3. Our World in Data, “Global average life expectancy has more than doubled since 1900,” Feb. 4, 2025 — https://ourworldindata.org/data-insights/global-average-life-expectancy-has-more-than-doubled-since-1900

  4. World Bank, “Poverty and Inequality Update,” Fall 2025 ($3-a-day line, 2021 international dollars; 2025 = estimate) — https://thedocs.worldbank.org/en/doc/229ff18129687a785f08af7cfb28e5e1-0350012025/original/WBG-Poverty-and-Inequality-Update-Fall-2025.pdf

  5. CDC, “1918 Pandemic” (at least 50 million deaths, estimate) — https://archive.cdc.gov/www_cdc_gov/flu/pandemic-resources/1918-commemoration/1918-pandemic-history.htm

  6. BEA, historical GDP estimates 1929–1999, 2000 (real GDP: about −26% from 1929 to 1933, 1929 level regained in 1936) — https://www.bea.gov/news/2000/gross-domestic-product-fourth-quarter-1999-final-estimate-revised-estimates-1929-99

  7. World Bank, “Rebuilding economies after COVID-19,” 2023 (world GDP −3.1% in 2020) — https://blogs.worldbank.org/en/opendata/rebuilding-economies-after-covid-19-will-countries-recover · WHO, May 24, 2024 (global life expectancy −1.8 years between 2019 and 2021) — https://www.who.int/news/item/24-05-2024-Covid-19-eliminated-a-decade-of-progress-in-global-level-of-life-expectancy

  8. OpenAI, “How People Use ChatGPT,” 2025 (700 million weekly users in July 2025) — https://cdn.openai.com/pdf/a253471f-8260-40c6-a2cc-aa93fe9f142e/economic-research-chatgpt-usage-paper.pdf

  9. World Bank, “Global Recessions,” 2022 (global recessions in 1975, 1982, 1991, 2009 and 2020) — https://documents1.worldbank.org/curated/en/099031409142266105/pdf/IDU01ed08ada0dcb1048f809c390d3ba1acab4e2.pdf

  10. Our World in Data, GDP per capita, Maddison Project Database 2023 (2022: $41,323 in Western Europe, $3,437 in sub-Saharan Africa, 2011 international dollars) — https://ourworldindata.org/grapher/gdp-per-capita-maddison-project-database

  11. Perez, Technological Revolutions and Financial Capital, 2002, ch. 4 — https://carlotaperez.org/wp-content/downloads/new-book/blog/the-second-machine-age/TRFCChapter4.pdf · ch. 7 (the turning point: the bubble bursts, then a recession whose length depends on the social and political capacity to reform institutions) — https://carlotaperez.org/wp-content/downloads/books/btr-en/PEREZ_TRFC_Ch%207.pdf · C. Perez research project, undated (“The global economy is at a turning point”; microelectronics and computerisation from the 1970s, the internet and globalisation from the mid-1990s, AI from the 2020s) — https://carlotaperez.org/research-project/

  12. International AI Safety Report 2026, Feb. 3, 2026 (announced investments, not spent) — https://internationalaisafetyreport.org/publication/international-ai-safety-report-2026 · Eurostat, Dec. 11, 2025 (enterprises with 10+ employees using AI: 20.0% in 2025, 13.5% in 2024) — https://ec.europa.eu/eurostat/web/products-eurostat-news/w/ddn-20251211-2

  13. Andrej Karpathy on X, Oct. 11, 2026 (his own estimates) — https://x.com/karpathy/status/2109361546505966046

  14. Paul A. David, “The Dynamo and the Computer: An Historical Perspective on the Modern Productivity Paradox,” American Economic Review, Papers and Proceedings, May 1990 (factory productivity effect in the early 1920s, some forty years after the first central power stations) — https://www.jstor.org/stable/2006600

  15. Our World in Data, “Life expectancy is returning to pre-pandemic levels,” Oct. 28, 2024 — https://ourworldindata.org/data-insights/life-expectancy-is-returning-to-pre-pandemic-levels

  16. Allen, “Engels’ pause: Technical change, capital accumulation, and inequality in the British industrial revolution,” Explorations in Economic History, 2009 (output per worker +46%, real wages +12%, 1780–1840, on Feinstein’s series) — https://gabriel-zucman.eu/files/teaching/Allen09.pdf · Crafts, Oxford Economic Papers, 2022 (alternative estimates more favorable to wages) — https://academic.oup.com/oep/article/74/1/1/6145336

  17. Acemoglu and Johnson, Power and Progress, 2023, authors’ overview (technical progress does not by itself produce shared prosperity; it takes choices and countervailing power) — https://shapingwork.mit.edu/power-and-progress/faqs/ · Nobel Prize in Economic Sciences 2024 (Acemoglu, Johnson, Robinson), Oct. 14, 2024 — https://www.nobelprize.org/prizes/economic-sciences/2024/press-release/

End of the series · AI & JobsAll series →