After 28 years spent literally or figuratively in Silicon Valley, I’ve grown increasingly concerned about AI’s potential to cause widespread, permanent unemployment. So I clambered out of the valley and into the “Ivory Tower” to share my fears with Israel’s leading economists. Turns out, the view from the Tower is wildly different from the Valley. Not a good thing.
The view from the Valley is that AI will achieve human-level intelligence within a few years, leading to rising unemployment.
Leopold Aschenbrenner, a former superintelligence researcher at OpenAI, says: “We are building machines that can think and reason. By 2025/26, these machines will outpace many college graduates. By the end of the decade, they will be smarter than you or I; we will have superintelligence, in the true sense of the word... That doesn’t require believing in sci-fi; it just requires believing in straight lines on a graph.”
Avital Balwit, chief of staff at Anthropic, says: “I am 25. These next three years might be the last few years that I work. I stand at the edge of a technological development that seems likely, should it arrive, to end employment as I know it.”
Consequently, tech rivals like Sam Altman, Mark Zuckerberg, and Elon Musk agree on the need to prepare for permanent mass unemployment with universal basic income.The view from the Tower is far more tranquil. The OECD’s report on “The Future of Work” concludes that “Overall, the total number of jobs is not likely to decrease,” while The World Economic Forum predicts “a net increase of 58 million jobs.” I heard much the same from Israel’s think tanks and policy institutes.
Who’s right? I hope it’s the economists, but I’d wager on the technologists for three reasons:
1. “Change is the only constant” (Heraclitus)
The Great Crash of 1929 saw 90% of the stock market’s value vanish. A few days prior, the prominent economist, Irving Fisher, pronounced that “stock prices have reached what looks like a permanently high plateau.” Economists have missed every crash since, prompting the IMF to conclude that economists are “notoriously poor at spotting a crisis coming”, and that there is “little evidence… that forecasts at horizons of two to five years contain much predictive content.”
Add technology to the mix, and the economists’ record goes from “notoriously poor” to comical. McKinsey, for example, pronounced that “mobile phones will never be a mass market,” while Nobel-winning economist Paul Krugman predicted the Internet’s impact would be “no greater than the fax machine’s.” In retrospect, Krugman conceded that “most macroeconomics of the last 30 years was spectacularly useless at best and positively harmful at worst”.
In contrast, technologists have an impressive record in predicting key milestones decades in advance. Writing in the ’80s and ’90s, computer scientist Ray Kurzweil accurately forecasted to within a couple of years the arrival of the Internet, smartphones, voice recognition, self-driving cars, and virtual reality. In 1990, he predicted that human-level AI would arrive in 2029, a prognostication that has since catapulted from preposterous to prescient.
The difference? The economy is governed by the butterfly effect, while technology is governed by Moore’s Law, which posits that computational power doubles every two years.
Kurzweil calculated roughly how much compute is needed for each milestone he envisaged, and predicted their realization at the point where these needs intersect with the exponential progression of Moore’s Law. His track record isn’t perfect, but no economist can hold a candle to it.
2. The boy who cried wolf
The second reason is that I find the economists’ arguments unconvincing. One explanation they offer for their equanimity is that, despite the rise of AI, unemployment has not risen at all. Yet nobody expected generative AI to move the macroeconomic needle so quickly; and in smaller, bellwether sectors, the needle is buried in the red. Freelance job boards, for example, have seen massive drops in demand for writers, web developers, graphic designers, and engineers.
More importantly, when a macroeconomic signal does emerge, I expect it to show that AI augments people rather than making them dispensable – right up to the point where it dispenses with them. By way of analogy, consider the story of Bob, a mediocre manager. In Act 1, Bob hires Sam, a wunderkind, who boosts the quality and quantity of Bob’s deliverables. The big boss is happy. In Act 2, Sam has learned the ropes and is able to fly solo. Bob looks expensive and incompetent by comparison. In Act 3, Bob gets canned. The End.
The second explanation offered for their poise is that we’ve seen this movie before and it has a happy ending. There’s full employment today even though 99% of the pre-industrial jobs have vanished. Stay calm and carry on.
But, unlike the industrial revolution, where machines replaced our brawn and we found jobs using our brains, today’s machines are set to outperform our brains. What part of our being will we use to earn a living once that happens?!
Oh, and the industrial revolution triggered a century of catastrophic hardship, including mass displacement of skilled workers, and a rush for raw materials that fueled colonialism and wars that claimed tens of millions of lives. Not a movie we want to take our kids to.
3. Two thumbs down
AI is approaching human-level performance across the spectrum of intellectual endeavors. At its current rate of progress, AI will soon project onto your screen a talking-head that will shape-shift to be your lawyer, graphic designer, doctor, software engineer – you name it. As a rule of thumb, therefore, we should assume that any job that can be done over Zoom today can be done by an AI tomorrow. I’ve encountered no credible counterargument to this.
Which leads to my second rule of thumb: employers will replace humans with AI whenever there’s a buck to be made. That is the true lesson from the industrial revolution. I’ve heard no credible counterargument to this either.
Silicon Valley has tunnel vision. Taken together, you see why, on balance, I’d wager on Silicon Valley’s predictions for what AI will soon do. But I’d never trust the Valley to tell society how to adapt or prepare. When it comes to the societal implications of its technologies, Silicon Valley is “spectacularly useless at best and positively harmful at worst.” Indeed, recent years have seen devastating unintended consequences of the Valley’s innovations, from skyrocketing teen suicides to the radicalization of our society.
Tech titans casually toss out slogans like “universal basic income” as though UBI is a panacea for the coming age. I favor UBI, but they seem oblivious to the monumental challenges it entails, including staggering costs, elusive sources, and complex second-order effects. Moreover, the societal problems born of mass unemployment won’t end with any universal income, let alone a basic one. We need the full engagement by the Ivory Tower, leveraging the expertise of economists, political scientists, and sociologists to navigate these intricate issues.
In the coming years, AI is likely to achieve superhuman intelligence and drive rising unemployment. To my knowledge, no one has articulated a convincing case for how such AI can coexist with full employment, and so we must prepare accordingly. Yet those who see the gathering storm are ill-equipped to prepare for it, while those who know how to prepare don’t see it coming.
Aesop’s Fable tells of two men, one blind, the other lame. Alone they can’t survive, so the lame man climbs onto the blind man’s back, and united they can navigate safely.
The moral is clear: we can’t rely solely on economists’ predictions or Silicon Valley’s optimism. We need technologists who understand AI’s potential, economists who can model its impacts, and policymakers who can implement solutions.
The clock is ticking. Our choices today will shape whether AI becomes humanity’s greatest achievement or, like the golem of Jewish lore, a force that turns on its master.
The writer is CEO and co-founder of Lemonade (NYSE: LMND), and chairman of the MOSAIC Policy Institute, whose mission it is to ensure that Artificial Intelligence benefits all of Israel’s society.