OpenAI CEO Sam Altman is a strong advocate for Universal Basic Income (UBI), and he spent three years conducting a large-scale experiment: In 2019, he selected 1,000 people in Texas and Illinois who were below the poverty line (with a household income of less than $29,000 per year) and gave them $1,000 per month for three years, while also selecting another 2,000 people with similar conditions and giving them $50 per month as a control group. The results of this study have now been published in a paper.
Before presenting the conclusions, it may be necessary to explain what Universal Basic Income is. It is a long-standing social welfare proposal that argues that national finances should "guarantee" a basic income for all citizens, without any conditions attached, to ensure that everyone can maintain a basic standard of living even if they do not work.
UBI has always been a utopian proposal, with many small-scale experiments and variations, but there has been a lack of large-scale research to validate its feasibility, especially the vision of its proponents that after ensuring a basic income for all citizens, human dignity will be further established, and people's creativity will be unleashed, creating greater economic benefits for society.
Sam Altman, as an AI industry practitioner, has a deeper understanding of AI replacing human jobs, so he has always been enthusiastic about establishing a UBI system. This project raised a total of $60 million, of which $14 million came from his own pocket, making it the largest cash assistance experiment in history.
So, what's the conclusion? The study found that this UBI test did not yield clear results. Isn't it a bit confusing? After investing so much money, is that all? Can you tell me whether this plan is feasible or not?
In fact, the paper is written in such a convoluted way that it is difficult to draw conclusions. Under the premise of being responsible to major donors like Sam Altman, the data and analysis presented in the paper are very difficult. In short, over the three years of cash distribution, there was no significant benefit or harm, making it difficult to make a judgment.
For convenience, I will refer to those who receive $1,000 per month as the "money group" and those who receive $50 per month as the "control group".
Compared to the control group, the money group increased their average monthly spending by $310, which was used only for food, transportation, and rent, while also reducing their work hours, averaging 5.2 hours less per month, resulting in a monthly income decrease of $125.
The researchers originally thought that cash assistance could allow people to have more freedom to find better job opportunities, but in reality, the money group chose to accept lower-paying, easier jobs, resulting in an overall decrease in job quality.
In other words, the assumption that UBI can enable people to create greater productivity with a more relaxed mindset does not exist in this experiment.
The health status of the money group did not improve significantly, with their mental state slightly improving in the first year, but then returning to their pre-payment state in the second and third years. The only thing that made the researchers optimistic was that the money group's alcohol consumption decreased by 20% and the number of days they needed painkillers decreased by 53%.
More surprisingly, excluding the cash assistance, the income growth of the money group was much lower than that of the control group - because the experiment overlapped with the pandemic period, the reduction in the money group's income per unit time and the overall income increase after the pandemic recovery are not contradictory - the money group was too complacent, highlighting the enthusiasm of the control group and proving the concern that cash assistance creates laziness.
However, the money group's participation in education and training was 14% higher than the control group, and the probability of moving out of poor communities was 11% higher, which somewhat shows that the UBI plan still has a positive impact on long-term skills investment, which is one of the few good news.
In any case, from the overall feedback of the sample, it is difficult to solve the structural poverty of poor families with a monthly subsidy of $1,000. You can say that the money is not enough, or that three years is too short, but there is no more massive funding cost to conduct a perfect experiment now. Sam Altman's project is the deepest one, and it may be because the conclusions are not very satisfactory that he and other major funders refused interviews from some media.
