Universal Basic Income Is Neither Universal Nor Basic

Artificial intelligence is going to put many out of jobs. But paying people not to work will only increase inequality and rancor.

Look out, Bangalore, she's coming for your jobs.

Photographer: Carl Court

As machine learning and robotics improve in the coming decades, hundreds of millions of jobs are likely to disappear, disrupting the economies and trade networks of the entire world. The Industrial Revolution created the urban working class, and much of the social and political history of the 20th century revolved around its problems. Similarly, the artificial intelligence revolution might create a new "unworking class," whose hopes and fears will shape the history of the 21st century.

Universal Basic Income

The social and economic models we have inherited from the previous century are inadequate for dealing with this new era. For example, socialism assumed that the working class was vital for the economy, and socialist thinkers tried to teach the proletariat how to translate its immense economic power into political clout. These teachings might become utterly irrelevant in coming decades, as the masses lose their economic value.

Indeed, some might argue that already, Brexit and Donald Trump’s presidential victory demonstrate an opposite trajectory. In 2016, many Brits and Americans who had lost their economic usefulness but retained some political power used the ballot box to revolt before it is too late. They revolt not against an economic elite that exploits them, but against an economic elite that doesn't need them anymore. It is far more frightening to be useless than to be exploited.

In order to cope with such unprecedented technological and economic disruptions, we probably need completely new models. One that is gaining increasing attention and popularity is universal basic income. UBI suggests that some institution -- most likely a government -- will tax the billionaires and corporations controlling the algorithms and robots, and use the money to provide every person with a stipend covering basic needs. The hope is that this will cushion the poor against job loss and economic dislocation, while protecting the rich from populist rage.

Not everybody agrees that UBI will be necessary. Fears that automation will create massive unemployment go back to the 19th century, and so far they have never materialized. In the 20th century, for every job lost to a tractor or a computer at least one new job was created, and in the 21st century automation has so far caused only moderate job losses. But there are good reasons to think that this time it is different, and that machine learning is a real game-changer. The experts who cry “job loss!” are a bit like the boy who cried wolf. In the end, the wolf really came.

Humans have basically two types of skills -- physical and cognitive. In the past, machines competed with humans mainly in raw physical abilities. Humans always had an immense cognitive edge over machines. Hence, as manual jobs in agriculture and industry were automated, new service jobs emerged that required the kind of brainpower only humans possessed. Now AI is beginning to outperform humans in more and more cognitive skills, and we don’t know of any third field of activity where humans retain a secure edge.

Of course, some new human jobs will develop in the 21st century, be it in engineering software or teaching yoga. Yet these will demand high levels of expertise and creativity, and will therefore not solve the problems of unemployed, unskilled laborers.

During previous waves of automation, people could usually switch from one low-skill job to another. In 1920, a farm worker laid off because of the mechanization of agriculture could find a new job in a factory producing tractors. In 1980, an unemployed factory worker could start working as a cashier in a supermarket. Such occupational changes were feasible, because the move from the farm to the factory and from the factory to the supermarket required only limited retraining.

But in 2040, a cashier or textile worker losing a job to an AI machine will hardly be able to start working as a software engineer or a yoga teacher. They will not have the necessary skills.

Proponents of UBI hope to solve that problem. Freed of economic worries, the unemployed could just forget about work, and devote themselves to their families, hobbies and community activities, and find meaning in sports, arts, religion or meditation.

Yet the formula of universal basic income suffers from several problems. In particular, it is unclear what “universal” and “basic” mean.

When people speak about universal basic income they usually mean national basic income. For example, both Elon Musk and former President Barack Obama have spoken about the need to consider some kinds of UBI schemes. But when Musk said that “There’s a pretty good chance we end up with a universal basic income … due to automation,” and when Obama said that “whether a universal income is the right model … that’s a debate that we’ll be having over the next 10 or 20 years,” it is unclear who “we” are. The American people? The human race?

Hitherto, all UBI initiatives were strictly national or municipal. In January, Finland began a two-year experiment, providing 2,000 unemployed Finns with $630 a month, irrespective of whether they find work or not. Similar projects are underway in Ontario, Holland and Livorno, Italy. Last year, Switzerland held a referendum on instituting a national basic income scheme, but voters rejected the idea.

In the U.S., Representative Ro Khanna, a California Democrat, proposes to greatly expand the Earned Income Tax Credit program, boosting the income of poor Americans by about $1 trillion. Though the plan does not promise any stipends to the unemployed, it is seen as a first step towards instituting national basic income.

The problem with such national and municipal schemes, however, is that the main victims of automation may not live in Finland, Amsterdam or the U.S. Globalization has made people in one country dependent on markets in other countries, but automatization might unravel large parts of this global trade network with disastrous consequences for the weakest links.

In the 20th century, developing countries made economic progress mainly by exporting raw materials or by selling the cheap labor of their workers and service personnel. Today, millions of Bangladeshis make a living by producing shirts that are sold to customers in the U.S., while people in Bangalore, India, earn their keep answering the complaints of American customers.

Yet with the rise of AI, robots and 3-D printers, cheap labor will become far less important, and demand for raw materials might also drop. Instead of manufacturing a shirt in Dhaka and shipping it all the way to New York, you could buy the shirt’s code online from Amazon and print it in Manhattan. Zara and Prada stores could be replaced by 3-D printing centers, and some people might even have such printers at home.

Simultaneously, instead of calling customer services in Bangalore to complain about your printer, you could talk with an AI representative in the Google Cloud. The newly unemployed workers and call center operators in Dhaka and Bangalore don’t have the education necessary to switch to designing fashionable shirts or writing computer code -- so how will they survive?

Under this scenario, the revenue that previously flowed to South Asia will now fill the coffers of a few tech giants in California, leading to huge strain on developing economies. American voters might conceivably agree that taxes paid by Inc. and Alphabet Inc. be used to give stipends to unemployed coal miners in Pennsylvania and jobless taxi-drivers in New York. However, does anyone think American voters would also agree that part of these taxes should be sent to Bangladesh to cover the basic needs of the unemployed masses there?

Another major difficulty is that there is no accepted definition for "basic" needs. From a purely biological perspective, the only thing a Homo sapiens needs for survival is about 2,500 calories of food per day. Over and above this biological poverty line, every culture in history defined additional basic needs, which change over time.

In Medieval Europe, access to church services was seen as even more important than food, because it took care of your eternal soul rather than of your ephemeral body. In today’s Europe, decent education and health care services are considered basic human needs, and some argue that even access to the internet is now essential for every man, woman and child.

So if in 2050 the United World Government agrees to tax Google, Amazon, Baidu Inc. and Tencent Holdings Ltd. in order to provide a basic income for every human being on earth, from Dhaka to Detroit, how will it define “basic”?

For example, will universal basic income cover education? And if so, what would these services include: just reading and writing, or also composing computer code? Just six years of elementary school, or everything up to Ph.D.?

And what about health care? If by 2050 medical advances make it possible to slow down aging processes and significantly extend human lifespans, will the new treatments be available to all 10 billion humans on the planet, or just to a few billionaires? If biotechnology enables parents to "upgrade" their children, would this be considered a basic human need, or would we see humankind splitting into different biological castes, with rich super-humans enjoying abilities that far surpass those of poor Homo sapiens?

Whichever way you choose to define basic human needs, once you provide them to everyone free of charge, they will be taken for granted, and then fierce social competitions and political struggles will focus on non-basic luxuries -- be they fancy self-driving cars, access to virtual-reality parks,  or enhanced bioengineered bodies. Yet if the unemployed masses command no economic assets, it is hard to see how they could ever hope to obtain such luxuries. Consequently, the gap between the rich (Tencent managers and Google shareholders) and the poor (those dependent on universal basic income) might become bigger and more rigid than ever.

Hence, even if universal basic income means that poor people in 2050 will enjoy much better medical care and education than today, they might still feel that the system is rigged against them, that the government serves only the super-rich, and that the future will be even worse for them and their children.

People usually compare themselves to their more fortunate contemporaries rather than to their ill-fated ancestors. If in 2017 you tell a poor American in an impoverished Detroit neighborhood that she has access to much better health care than her great-grandparents did in the age before antibiotics, it is unlikely to cheer her up. Indeed, such talk will sound terribly smug and condescending. "Why should I compare myself to nineteenth-century peasants?" she might retort. "I want to live like the rich people on television, or at least like the folks in the affluent suburbs."

Similarly, if in 2050 you tell the useless class that they enjoy better health care than in 2017, it might be very cold comfort to them, because they would be comparing themselves to the upgraded super-humans who dominate the world.

Modern communication systems make such comparisons almost inevitable. A man living in a small village 5,000 years ago measured himself against the other 50 men in the settlement. Compared to them, he probably looked pretty hot. Today, a man living in a small village compares himself to the 50 most gorgeous hunks on the planet, whom he sees everyday on TV screens and giant billboards. Our modern villager is likely to be far less happy with the way he looks.  Will universal basic income include plastic surgery for everyone?

Homo sapiens is just not built for satisfaction. Human happiness depends less on objective conditions and more on our own expectations. Expectations, however, tend to adapt to conditions, including to the condition of other people. When things improve, expectations balloon, and consequently, even dramatic improvements in conditions might leave us as dissatisfied as before.

If universal basic income is aimed to improve the objective conditions of the average person in 2050, it has a fair chance of succeeding. But if it is aimed to make people subjectively more satisfied with their lot in order to prevent social discontent, it is likely to fail.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

    To contact the author of this story:
    Yuval Noah Harari at

    To contact the editor responsible for this story:
    Tobin Harshaw at

    Before it's here, it's on the Bloomberg Terminal.