Author: Hao Boyang
Editor: Guo Xiaojing
After ChatGPT became popular, some people began to worry about losing their jobs.
When the AGI era truly arrives, AI will be able to do everything that humans can do, and do it faster and better than humans. So how can humans compete with them for jobs?
One of the answers given by politicians and leaders of large companies is the UBI plan, or unconditional basic income. It assumes that after the huge increase in productivity brought by AI, people will no longer have to work, and the government and companies will give everyone unconditional money to sustain their lives. This sounds good, but can it really bring benefits to people's lives? Politicians and academics have been arguing about this.
For years, OpenAI CEO Sam Altman has publicly expressed concerns that basic income will become necessary as artificial intelligence replaces traditional jobs and a few people accumulate huge wealth. "If public policy fails to adjust accordingly, most people will end up worse off than they are today," he wrote in 2021.
Olman is not alone. From Elon Musk to Geoffrey Hinton, known as the "father of artificial intelligence", many heavyweights in the technology industry believe that artificial intelligence will trigger a wave of technological unemployment, and basic income guarantees will become necessary to maintain our livelihoods.
Not only did Sam Altman express his concerns, he also funded a research institute called OpenResearch, which has teamed up with the U.S. government to give people unconditional money and observe what changes will occur in their lives.
The researchers selected 3,000 Americans with an average annual household income of about $30,000 and divided them into two groups, 1,000 of whom received $1,000 per month and the other 2,000 as a control group who received $50 per month to see what the effect would be. The program lasted for three years, from November 2020 to October 2023. That is to say, 1,000 of them received a total of $36,000 in three years, which is about 260,000 yuan in additional income.
(Grouping method)
Today, they released their preliminary research results.
How do recipients spend their money?
What do these people who receive $1,000 a month do with the money?
Not everyone will increase their spending. The researchers divided all recipients into three income levels: those with an annual income of less than $15,000 are low-income, those with an annual income of $15,000-60,000 are middle-income, and those with an annual income of more than $60,000 are high-income. Among them, the middle-income group had the largest additional expenditure, spending about half of the subsidy, while the poor group only spent 30%, and the rich group had basically no change. This is not that they did not spend the money they were given, but that the total expenditure did not change. In summary, the savings of all subjects increased by 25%.
(Specific costs for different groups)
Those who received $1,000 first met some of their basic needs. Those with the lowest incomes seized the opportunity to move out of old houses where four generations may have lived together and start living independently. Others will spend the money on basic needs such as food and transportation. According to Openresarch, about 62 cents of every transferred dollar is used for basic expenses such as food, rent and transportation, 22 cents is used for leisure, and 3 cents is used to increase borrowing because recipients take on more car loans and mortgages.
(Subjects’ expenses by category)
But what is surprising is that the fourth expenditure after just needs is actually giving an average of US$22 more to others every month, such as helping relatives in need or giving gifts to friends. And it's the largest increase of any type of spending. This effect is particularly pronounced among recipients from households with lower initial incomes and is more common among people of color. In the examples given by the researchers, they would repay relatives and friends who had helped them in the past but were now in trouble; some would even provide support to homeless friends in foreign countries.
(Subjects’ explanation of their behavior of supporting others)
Being poor does not mean being selfish.
The next most expensive items are children's expenses, household bills and medical care. In terms of luxury spending, such as clothing, alcohol and tobacco, and vacations, they all rank last, accounting for less than 10%.
(From a percentage point of view, the subjects' spending direction)
How did this cash change their lives?
Besides spending the money directly on things, how has this money changed their lifestyle?
First of all, it’s very simple. When I have money, I work less.
The study showed that the subjects all worked fewer hours, but they did not quit. Their weekly work hours were reduced by only 1.3 hours, and their income was reduced by 4-5%. Interestingly, the partners of the experimental participants also seemed to reduce their labor supply by the same amount. One person gets the money, and two people enjoy it.
(Comparison of working hours, red is the experimental group)
(Overall income changes, red is the experimental group)
They mainly use the time saved for socializing, caring for others and leisure.
The change in working hours is actually more obvious in the high-income group. They tend to reduce their working hours or the number of jobs they hold, rather than using the funds to increase spending. Low-income recipients can only use the money to improve their lives and dare not give up their jobs.
(Biggest change in monthly time allocation for the recipient group)
The researchers further pondered, after these subjects got more money and had more space, would they think more about life, work environment, and even improve their education? First, they observed whether the subjects went to find a better job, but because of their skill limitations, they were mainly engaged in previous level work. But now they will take an extra month off during unemployment, and will apply for fewer jobs, so they have more space to consider some better positions. They are even more likely to look for jobs based on their interests and meaning. Among job seekers, the probability that respondents believe that interesting or meaningful work is a necessary condition for them to accept any position is 5.5 percentage points higher than others.
(Interview with the recipient on job choice)
What's more, recipients are on average more likely (12% more likely) to want to plan for big expenses, that is, plan for the future, than they were previously just trying to get by. This shows that the improved environment has given them more expectations and motivation for the future. By the third year of the program, their participation in education or vocational training was significantly higher (3.3 percentage points), and recipients were also significantly more likely to report planning to further their education (6 percentage points).
(Beneficiaries’ comments on future plans)
Although they did not change jobs, some people did use the money to start a business. By the third year of the subsidy, 43% of black people began to plan to start a business, which was 9% more than the control group. There were also 5% more female recipients who intended to start a business than the control group.
(Changes in the entrepreneurial tendencies of recipients)
All this shows that due to their surplus wealth, the poor have gained greater initiative.
In addition to having more hope of living, this extra income also makes them more willing to go to the doctor. Although they spent far less on medical care than was needed, recipients had a 26% increase in hospitalizations last year compared with the control group. The probability of visiting outpatient clinics and dentists has also increased significantly.
(Changes in the recipient’s medical condition)
Perhaps because they have more hope for life, the rate of alcoholism in this group has dropped by 20%, and drug abuse has dropped by 53%. So the mountain people who are addicted to alcohol and drugs in "Hillbilly Elegy" may be only $1,000 a month away from a normal life.
Although the statistical impact report of UBI on crime has not yet been released, the reduction in drug and alcohol use is believed to lead to a certain degree of reduction in crime rates.
(The beneficiaries talk about the changes in their living conditions)
However, for most people, $1,000 is not enough. They need it to support their families. So while health spending increased, it was limited. As a result, the study found that cash had no effect on physical health indicators, self-reported access to health care, or health-promoting behaviors such as exercise and sleep.
(Changes in various indicators related to the recipients’ health)
In the stories of our recipients, we understand that they have multiple needs. Out of necessity, they may prioritize needs other than health—such as housing, food, transportation, and childcare expenses. Especially for low-income people, they face countless other pressures on their time, finances, and emotional bandwidth.
(Recipient's consideration of changes in medical circumstances)
So, is UBI reliable?
In the past, there have been four main arguments among opponents of UBI implementation.
First, it's too expensive. Indeed, it cost $14 million over three years just to meet the needs of this experiment covering 3,000 people. But if AI can really replace humans on a large scale, is this cost expensive relative to the value and wealth they generate? According to research by the International Monetary Fund, artificial intelligence may affect about 60% of jobs in developed economies, about half of which can be automated, which may lead to reduced recruitment and lower wages. Another study shows that by 2030, nearly 12 million American workers may need to change jobs. AI companies are promising us a more prosperous era. In Goldman Sachs' research, the US GPT alone can achieve a growth of more than one trillion. According to the UK welfare department's estimate, to achieve UBI, it is necessary to transfer 120 billion pounds of taxes to the welfare system. Then it only takes a small part of this part of GDP growth to subsidize the workers who are most affected. Is this expensive?
Don’t forget that these people are also working and paying taxes. Although their taxes are far less than the additional income they receive, for AI+artificial intelligence, the overall output is a net increase.
The second and most frequently mentioned point is that supporting the poor is equivalent to supporting lazy people. Only the possibility of falling into poverty can motivate people to become rich. Those who oppose UBI often believe that unconditional monetary assistance will encourage people's complacency and hinder them from setting and pursuing goals. However, the results of the study show that the recipients have not given up their work, but have only gained reasonable leisure time. Moreover, they have enhanced their planning awareness, thinking about the future, budget management ability and educational ambitions. In other words, they are more motivated. This is also the most important conclusion proved in this survey.
Third, UBI may reduce reform or even abolish the existing welfare system. This makes the poor more vulnerable. This reason may be even weaker. First, the two can actually go hand in hand. Second, trying to design a welfare system that meets so many different needs and desires in a centrally planned way is neither realistic nor feasible. Therefore, why not give people their own initiative and let them distribute more wealth themselves?
Fourth, it is unfair to taxpayers to distribute money uniformly, because it is taxpayers' money that is distributed. This actually assumes that the recipients are not taxpayers. Experiments have shown that the recipients are still willing to work, so they are also taxpayers. In addition, companies use AI to replace manpower and obtain higher profit margins and greater benefits. This part of the benefits enjoyed by technological progress can be used to compensate those who are difficult to be radiated by technological progress and are left behind. This is reasonable. After all, history is not created by individuals or single companies, and neither is technology.
So, what’s so unreliable about UBI?