AI is coming for our jobs. AI titans say universal basic income will save us. Will it?

When asked in a recent interview how the productivity gains of artificial intelligence can be evenly distributed, Geoffrey Hinton, computer scientist and ‘Godfather of AI’, replied in one word: “Socialism.” 

Even AI fat-cats have hinted at socialized returns. 

“I think society will very quickly say, ‘Okay, we’ve got to have some new economic model where we share that and distribute that to people,’” OpenAI CEO Sam Altman told podcaster Theo Von in July. “I still am kind of excited about universal basic income where you just give everybody money,” he added.

"I don’t think we’re going to have a choice,” Tesla CEO Elon Musk in 2017 on the question of UBI.

Listen closely and one can hear the sound of “I told you so” echoing from Karl Marx’s grave. The philosopher saw an inherent contradiction at the heart of capitalism, between the drive for efficiency and the resulting inability of workers to buy the goods produced. “The production of too many useful things results in too many useless people,” the philosopher wrote in Capital Volume III (1894), leading to crisis. What emerges: socialism.

So, in the AI economy, can we expect our Big Tech overlords to share the means of production? Maybe. Will universal basic income, or UBI, redistribute wealth? Unlikely. While it may be an effective rebuttal to concerns about mass unemployment, the economics are thorny and the outcomes are flawed. Here's why.

A 'radical policy'

UBI is considered a "radical policy" whereby cash is paid to all, without means testing, with no strings attached, and at a “sufficiently high level to enable a life free from economic insecurity,” according to Juliana Uhuru Bidadanure, professor of philosophy at Stanford University.

This concept isn’t new. In 1797, Thomas Paine proposed that a lump sum be granted to all citizens paid for by a tax on inherited land. The idea returned during the Civil Rights movement, pitched as the way to reverse entrenched inequality. "The time has come for us to civilize ourselves by the total, direct, and immediate abolition of poverty,” by means of the “guaranteed income,” Martin Luther King Jr. wrote in 1967. 

Amid rising fears about the impact of automation on labor, it has been revived again over the last decade. This time around, it's moving from a utopian ideal to an empirically studied possibility.

By 2030, 30% of current U.S. jobs could be automated, one McKinsey report projects. Goldman Sachs predicts 6-7% of the workforce will be displaced, which would take unemployment to around 12%, factoring in the current rate. To put that into perspective: during the 2007-2009 Great Recession, unemployment didn’t climb above 10%. Others foresee something far more radical in the long-term. “There will come a point where no job is needed,” Musk said in 2023.

Martin Ford, futurist and author of "Rise of the Robots," has argued for more than 15 years that automation will require unprecedented measures. In order for the AI economy to succeed, society must "find a mechanism to get purchasing power into the hands of consumers,"says Ford. "It doesn't matter if the stuff is all made by machines, that's fine, but people have to have money to buy it." The answer, he says, is UBI.

How much would be paid?

It is estimated that UBI of $10,000 annually would cost $3 trillion each year, equivalent to about three-fourths of the federal budget. Musk envisions something far more decadent. “There will be universal high income (not merely basic income). Everyone will have the best medical care, food, home, transport and everything else,” he said on X last month when asked how those made jobless from robots will sustain their lives. “Sustainable abundance,” he added.

But let's assume the policy starts with $10,000. Would this be enough to counterbalance the more modest projection that 6-7% of the workforce will be displaced?

It'll come down to who those workers are. Professions most at risk are computer programmers, legal assistants, copyeditors, accountants, and auditors, according to the Goldman Sachs report. “One of the things that sets AI apart is its ability to impact high-skilled jobs,” the IMF said last year. 

And it just so happens that these workers — white-collar professionals — play an inflated role in economic activity. The top 10% of U.S. earners — households earning more than $250,000 — accounted for nearly half of total consumer spending during the second quarter of this year, according to Moody’s Analytics, a record high since data collection began in 1989. 

“If you hit a good chunk of that top 10%, that has a massive impact on the overall economy,” says Ford. 

UBI of $10,000 wouldn’t begin to scratch the surface of the former incomes of the top 10%. Even considering that $10,000 will be evenly distributed across the population, the sum may fail to keep consumer spending growing at the rate that ultra-efficient AI-run businesses were producing goods and services.

How would it be funded?

AI will increase global GDP by $7 trillion — or 7% — over 10 years, predicts Goldman Sachs. Or, if you go with McKinsey’s estimate, it will grow between $17.1 and $25.6 trillion annually. The idea is UBI would be funded by taxing this swelling capital, presumably via corporation tax. This is rational: In the AI economy, capital will be doing more of what labor used to, so should be taxed more. But this assumes two things.

For one, convincing policymakers to enact tax-funded UBI would be “extraordinarily difficult,” Ford admits. It’s certainly hard to picture, given Congress just passed President Donald Trump’s One Big Beautiful Bill, which gives businesses around $1.1 trillion in tax breaks over the next decade. 

Moreover, this year has also seen Silicon Valley yield greater power in Washington. A.I. titans like Musk “not only own the technology but also effectively own a lot of politicians, Trump included,” writes the New Yorker’s John Cassidy. Betting on the likes of OpenAI, Meta, and Google to support the state seizing their returns is hard to conceive of in the current climate.

Secondly, what happens if mass joblessness strikes before AI has generated the trillions of dollars analysts have predicted?

In theory, the threshold AI must meet in order to automate white-collar work could be lower than the threshold it must reach in order to boost economic growth. Ford contends that the former is lower than the latter. 

“My concern is that unemployment starts first, and that's going to have a massive impact on the economy,” Ford says. 

And that unemployment won’t be historically comparable, he argues. During the COVID pandemic, when U.S. unemployment hit almost 15%, white collar workers — dubbed the “laptop class” — were able to work from home, while lower-income frontline workers, such as those in hospitality and retail, faced the brunt. 

AI will be “sort of the opposite" of the pandemic, says Ford, and for that reason, the impact could be “much more dramatic.”

And as workers essentially watch their professions go extinct, that could in turn worsen consumer confidence. “I wouldn't underestimate the psychological impact of that, and therefore the impact on the economy,” says Ford. 

Should we achieve innovations like superintelligence — AI that surpasses even the top human minds — then the “only constraint on the economy would be the laws of physics,” according to The Economist. 

Yet if unemployment triggers a recession before we reach it, “could be the thing that actually dampens down progress in AI, and means that maybe we don't get to that point where we can enjoy all those benefits,” says Ford.

'Bread and circuses'

But let’s say obstacles of lawmakers or recessions don’t manifest. There are also issues inherent to guaranteed income as a policy.

The largest randomized basic income experiment to date was held between 2020 and 2023 by OpenResearch, a nonprofit founded by Altman in 2015. One thousand lower-income households were given $1,000 per month, no strings attached. The study found that although recipients primarily used the money on their basic needs, leading to greater agency and financial stability, “the average effects across outcomes were limited,” says Elizabeth Rhodes, research director of the project. For instance, no significant improvements to physical health or academic performance were reported.

When it comes to the question of distributing AI-funded UBI across the entire population, the study suggests this may alleviate baseline poverty, as recipients did not use the unconditional cash recklessly. At the same time, it suggests that UBI would by no means be a socio-economic equalizer. As Rhodes put it: “Cash alone cannot address every challenge or serve as a singular solution.” 

One could even argue that UBI could worsen inequality in the AI economy. For one, the results show that income didn’t equate to better outcomes. More urgently, however, one must consider how unevenly wealth is distributed. AI capitalism would see assets surge, buoyed by the meteoric Magnificent Seven, while those without assets would have fewer options of building a career if entry-level white collar work is obliterated. Preventing the wealth gap from widening would require more aggressive income tax on capital gains, rather than on businesses alone.

But basic economic theory suggests taxes on income discourage work and investment. What’s more, “governments should avoid transfers to the same people from whom they collect revenue, but that is precisely what a UBI would do,” Daron Acemoglu, economics professor at MIT and Nobel laureate, wrote in 2019. 

Acemoglu also argues that replacing all other social welfare with UBI is a “terrible idea” as it could mean, for example, those with severe health conditions are denied adequate care in order to divert that money to billionaires. 

UBI, he argues, has the hallmarks of “bread and circuses” used by the Roman and Byzantine Empires: give handouts to placate the masses, rather than providing them with economic opportunities and political agency. "Though UBI makes for a good slogan, it is a poorly designed policy,” he adds. 

The end of expertise

Fellow MIT economist Professor David Autor has fallen short of explicitly condemning UBI, but argues that talk of mass joblessness is both defeatist and avoidable. It will create new, better jobs for America’s middle-class, middle-skill workforce and “reshape the value and nature of human expertise," he wrote in a paper published last year. AI has the potential to democratize expertise, allowing more workers to engage in “higher-stakes decision-making."

Earlier this month, when asked by the BBC about the fate of graduate-level jobs, Google’s president Ruth Porat cited one of Autor’s statistics: 60% of today’s jobs didn’t exist 80 years ago. “What we've seen throughout history is that new industries are created,” she added. 

Ford is skeptical of this optimism. In his view, AI can’t be compared to previous technological advances like the internet, because it is replacing our core competence. Hence, if innovations like superintelligence are achieved, it’s hard to see why businesses would use middle-skill workers for decision-making. 

The adage “no one is an expert if everyone is an expert” both captures the risk and promise of AI. One reading goes as such: It will take us beyond the expertise economy, where the best jobs are reserved for those with college degrees. Instead, more nuanced skills — judgment, leadership, innovative thinking — will grow in currency, ushering in new types of jobs. 

An alternative interpretation hints at something more ominous. That is, homo sapiens have handed over cognition — the thing that allowed them to rule over the animal kingdom and build civilization — to something that’s unburdened by corporeal fallibility. Should bots outsmart us, it feels somewhat inevitable that some version of UBI will follow. Left with a clunky, and arguably regressive, policy, may reveal the paradox of capitalism about which Marx cautioned.