Mass unemployment induced by AI would be unprecedented
Illustration: Katie Martin
Listen to this story
Your browser does not support the <audio> element.
A T NO TIME in polling history have Americans been less optimistic about their long-term employment prospects. The average person believes they have a 22% chance of losing their job in the next five years, according to one survey, a higher share than even during the global financial crisis of 2007-09. The cause of this gloom is artificial intelligence. Nearly one in five American workers recently told another pollster that AI or automation is “very” or “somewhat” likely to replace them.
It isn’t just average people who are alarmed. So are the leaders of the very AI companies causing the anxiety. Dario Amodei of Anthropic has warned that AI could push unemployment to 10-20%. Bill Gates, co-founder of Microsoft, said that in an AI world people will not be needed for “most things”. Sam Altman, boss of Open AI, has clocked that talking up the technology’s disruptive power is provoking a backlash, and now speaks of “tools to augment and elevate people, not entities to replace them”. But even he could not resist mentioning “disruption/significant transition as we switch to new jobs”.
Economists are, for a change, far less dismal. They are allergic to the “lump of labour fallacy” which treats the jobs market as static and zero-sum. If technology displaces workers from some occupations, they argue, it enriches others, who then spend their gains on goods and services that create new employment.
Chart: The Economist
The labour market certainly is not cracking yet. The share of the OECD ’s working-age population with a job keeps breaking records (see chart 1), unemployment across the club of mostly rich countries is just 5%, and America employs more people than ever in “ AI -exposed” industries like law. American graduates have been struggling since before Open AI launched Chat GPT in late 2022. Many economists foresee relatively little disruption ahead. Those at America’s Bureau of Labour Statistics think the country will add 5.2m jobs between 2024 and 2034, increasing total employment by 3%.
Advances in AI ’s capabilities could render current data, and extrapolations from this, obsolete. But if this happened, and AI really were to put millions of people out of work, it would be unprecedented in human history. Never have new technologies spread fast enough to make large numbers of people unemployed for a long time. Understanding why may shed light on how this time is—and is not—different.
Historical data suggest that technological diffusion always proceeds slowly. In a paper published in 2012 Robert Gordon of Northwestern University found that since 1300, growth of GDP per person at whatever was the world’s most sophisticated economy of its time has never exceeded about 2.5% a year. When other countries grew faster than this, they did so by catching up with a richer place that, almost by definition, sparked earlier wealth-creating technological progress. And the fact that growth at the frontier of innovation was slower meant that so was the pace of any job destruction.
Take farming. Although it has undergone monumental technological upheavals over the past millennium, farm employment has changed only slowly. The share of England’s labour force in agriculture has been falling steadily since the 16th century without ever collapsing suddenly. The recognisably modern tractor was invented in America at the start of the 20th century, and it took generations rather than years for the agricultural workforce to decline.
Even when job disruption is faster, workers need not suffer. In the middle of the 20th century the first computers, shipping containers and other wonders led Harold Wilson, a British prime minister, to describe the “white heat of technology” burning through Western economies. GDP per person in America, which had by then dislodged Britain as the world’s frontier economy, grew by 2.5% a year, the fastest ever for a leading economic power. The level of job disruption, as measured by the share of employment shifting between industries or occupations, was at times more than twice as high as it is today. Yet many people look back fondly on that era as a time of rising wages, widening opportunity and unpolarised politics.
One instance of technological change has become notorious: the Industrial Revolution in 19th-century Britain. According to some accounts, it was horribly disruptive to workers. James Watt’s inventions in the 1760-1780s made steam engines efficient enough to power factories. This led to a period of blistering economic growth that appeared to coincide with stagnation in inflation-adjusted wages. Between 1790 and 1840 these barely budged, even as capitalists earned vast profits.
Chart: The Economist
Today’s “thought leaders” in Silicon Valley often invoke this pause. It is associated with Friedrich Engels, a capitalist-heir-turned-communist who described it in “The Condition of the Working Class in England”, his account of Manchester’s slums in the 1840s. Recent scholarship, though, casts doubt on whether “Engels’ pause” is a useful blueprint for what AI may have in store for workers.
The composition of British employment saw little churn until the 1850s, and then only as much as it does today (see chart 2). Moreover, if technology destroyed jobs, it created plenty more. Between 1760 and 1860 the number of Britons in work ballooned from 4.5m to 12m. Unemployment generally remained modest (se chart 3).
Chart: The Economist
Wage growth was indeed slow during Engels’ pause—but no slower than in the half-century before it. This reflected slow productivity growth in the Industrial Revolution’s early years, itself a function of the gradual diffusion of Watt’s technological breakthroughs. By 1830 only about 160,000 horsepower was in use in the whole of Britain, equivalent to 1,000 typical modern cars. Given rapid population growth in the period, it is a “truly remarkable achievement” that workers’ purchasing power grew at all, as Sir Tony Wrigley, a late British demographer, put it. It looks even more remarkable if you adjust wages not by the consumer-price index, as historians tend to, but by the average price of domestically produced output, the “ GDP deflator” (see chart 4).
Chart: The Economist
The gap between the two measures of real wages illustrates a crucial point about the Industrial Revolution. The average employer paid workers reasonably fairly after selling his wares and deducting the cost of materials. He did not profit from exploiting his staff, as Engels supposed. The problem for labourers was less unfair pay than sharp rises in the cost of living. Food prices rose steadily, and sometimes soared, because of war and high tariffs on grain imports. The villains of the Industrial Revolution were politicians, not machines.
This puts a different gloss on the industrial unrest of the period. In the early 19th century textile workers revolted, destroying the power looms they thought would kill their craft. A few years later farm labourers smashed threshing machines across southern England. Historians link such unrest to technological disruption, yet strikes and wrecking are as old as time. In England, riots were less frequent in the early 1800s—the middle of Engels’ pause—than later in the century, when real wages were growing strongly. The Chartists, who secured suffrage and other rights for working men, did not gain ground until wage growth was unpaused in the 1840s.
Nicholas Crafts, an economic historian, summed it up neatly. The Industrial Revolution, he wrote, is “not a template” for “technological change that [boosts] productivity at the expense of a significant …decline in labour’s share of national income”. In short, those warning of AI -driven mass unemployment are predicting something that has never happened before.
That does not mean it can never happen at all. The first signs would be sharply rising productivity combined with weak real-wage growth in America, the world’s frontier economy. This would show up as an increase in GDP per person, above Mr Gordon’s ceiling of 2.5%, and a simultaneous jump in corporate profits, reflecting the gains from higher output flowing to capital, not labour. Another signal would be big job losses in lots of industries.
History holds a final lesson. If disruption is coming, it will show up in a recession. Downturns cleanse the economy of unproductive jobs. Companies must make radical changes to survive; weak firms go under; capital and labour moves to more productive ones. Almost all of America’s once-routine jobs have vanished during past downturns. Which ones vanish next time will offer a big hint. Until then everyone—including Messrs Amodei, Gates and Altman—will remain none the wiser about the shape of the AI world to come.■
논증 분석
유형: causal
핵심 주장
AI로 인한 대규모 실업은 역사상 전례가 없는 일이며, 과거 기술 혁명의 사례들은 기술이 대량 장기 실업을 초래하지 않았음을 보여준다.
논리구조
- 전제: 현재 미국인들의 고용 비관론은 역대 최고 수준이며, Dario Amodei, Bill Gates, Sam Altman 등 AI 업계 리더들조차 AI로 인한 실업 위기를 경고하고 있다.
- 반론: 경제학자들은 ‘노동량 고정 오류(lump of labour fallacy)‘를 경계하며, 기술이 일부 직업을 대체해도 새로운 수요와 일자리를 창출한다고 주장한다. 실제로 OECD 취업률은 역대 최고를 기록 중이며 미국의 AI 노출 산업 고용도 증가하고 있다.
- 논거: Robert Gordon의 2012년 연구에 따르면 1300년 이후 세계 최선진 경제의 1인당 GDP 성장률은 연 2.5%를 넘은 적이 없으며, 이는 기술 확산과 일자리 파괴 속도 모두 역사적으로 항상 느렸음을 의미한다.
- 논거: 농업의 경우 현대적 트랙터 발명 이후에도 농업 노동력이 감소하는 데 수십 년이 아닌 수 세대가 걸렸으며, 영국 농업 노동 비중은 16세기 이후 급격한 붕괴 없이 꾸준히 완만하게 감소했다.
- 논거: 20세기 중반 컴퓨터·컨테이너 등 기술 혁명기에 미국의 산업 간 고용 이동은 현재의 두 배 이상이었지만, 그 시대는 실질임금 상승과 기회 확대의 시기로 기억된다.
- 진단: ‘Engels’ pause’로 알려진 산업혁명기 영국의 실질임금 정체는 기계에 의한 착취가 아니라, 전쟁과 곡물 수입 관세로 인한 식품 물가 급등에서 비롯된 것이었다. 즉 산업혁명의 악당은 기계가 아니라 정치인이었다.
- 진단: 산업혁명 당시 영국의 고용 구성 변화는 미미했고, 1760~1860년 사이 취업자 수는 450만에서 1,200만으로 증가했으며 실업률도 대체로 낮게 유지되었다. Nicholas Crafts는 산업혁명이 노동 소득 비중 하락을 수반한 기술 변화의 ‘템플릿’이 아니라고 결론지었다.
- 결론: AI 주도 대량 실업을 경고하는 이들은 역사상 한 번도 일어난 적 없는 일을 예측하는 것이며, 그 첫 신호는 생산성 급등과 실질임금 정체의 동시 발생, 그리고 다수 산업에서의 대규모 일자리 감소일 것이다.
- 처방: 역사적으로 기술에 의한 일자리 파괴는 경기침체기에 가시화되었으므로, 다음 불황에서 어떤 직종이 사라지는지를 관찰하는 것이 AI 시대의 노동시장 변화를 판단하는 핵심 지표가 될 것이다.
결론
AI로 인한 대규모 실업은 역사적으로 전례가 없으며, 현재까지 그 징후도 없지만 만약 발생한다면 생산성 급등·실질임금 정체·경기침체기 대량 실직이라는 신호로 먼저 포착될 것이다.
For more expert analysis of the biggest stories in economics, finance and markets, sign up to Money Talks, our weekly subscriber-only newsletter.