The landmark study provides more data on the benefits of universal basic income, the OpenAI founder’s favored solution for a future in which AI takes everyone’s jobs.
By Sarah Emerson, Forbes Staff
For the past eight years, an experimental project financed by OpenAI’s billionaire cofounder and CEO Sam Altman has been quietly testing a utopian idea: What if everyone in the world were given free money, regularly, with no strings attached?
“Universal basic income” was one of the first concepts probed by OpenResearch, a moonshots lab linked to OpenAI that Altman has personally contributed tens of millions of dollars to in a crusade to shape a future he sees as inevitably disrupted by artificial intelligence. Now, the project is publishing the results of an extensive trial that cumulatively gave away $45 million to thousands of people across America, in what it has called “the most comprehensive study” ever on guaranteed income.
The study’s findings were shared today in two papers published by the National Bureau of Economic Research. They are the first of several that OpenResearch plans to release, detailing a three-year-long trial in which 3,000 participants in Texas and Illinois were chosen at random to receive a $1,000 or $50 monthly stipend. The point of the investigation was to learn how our lives might change if we were given a small unconditional allowance. Its initial findings reveal that people who received this money tended to spend it on basic needs, medical care and helping others. Upcoming papers will focus on subjects like children, mobility, crime and politics.
Throughout the trial, researchers collected data from phone and online surveys, interviews and time diaries, as well as third-party sources like educational records and credit reports. They also drew blood from willing participants to track changes in certain health biomarkers. Once their analysis is finished, the team hopes to de-identify and publicly share their data set. “Our goal is to just produce the data and make it available in whatever form works best for people, and as widely as possible,” OpenResearch director Elizabeth Rhodes told Forbes.
Theirs is hardly the first endeavor to measure the benefits of a guaranteed income, but the study by OpenResearch is on the larger end of several dozen pilot programs around the world. The biggest is a 12-year trial in Kenya that started in 2017 and is funded by the philanthropic organization GiveDirectly. Countries like the U.S. and Canada have also flirted with the concept. Since the 1980s, residents of Alaska have received annual payments generated by the state’s oil and gas royalties. And last year, California launched its first state-funded guaranteed income test, which will target former foster youth.
Karl Widerquist, a basic income historian and professor at Georgetown University in Qatar, said we’re currently experiencing a “third-wave basic income movement” after witnessing its popularity rise in fits and starts over many decades. He was contacted by OpenResearch a few years back to provide his thoughts on the trial, which had not yet begun, and told Forbes they chose “decent amounts” to study. Now, he wants the federal government to move forward on implementing basic income. “We have lots of data on what basic income can do. We just disagree about whether we want this to happen.”
Altman has said repeatedly that he sees universal basic income as a poverty solution, dating back to his time as the president of startup accelerator Y Combinator. In a blog post from almost a decade ago, he made a unique appeal to researchers. “We’d like to fund a study on basic income,” he wrote. “I’ve been intrigued by the idea for a while, and although there’s been a lot of discussion, there’s fairly little data about how it would work.” Recently, basic income has been evangelized by Silicon Valley technologists who see it as a salve for human joblessness caused by automation. “It’s going to be necessary,” Elon Musk claimed in 2017, as “there will be fewer and fewer jobs that a robot cannot do better.” (He changed his mind this year, saying “We won’t have universal basic income. We’ll have universal high income,” without explaining the difference.) Altman has called it an “obvious conclusion” to his prediction that “computers will replace effectively all manufacturing.”
Some technologists remain skeptical. Computer scientist and “godfather of virtual reality” Jaron Lanier maintains a friendly disagreement with Altman and others who’ve endorsed AI-subsidized social welfare. Lanier told Forbes that, in the attempt to create a more egalitarian society, basic income runs the risk of centralizing this flow of wealth. Assuming superintelligence is just beyond the horizon, “I’d like to see people become proud data providers in a new economy” as a way out of this plutocratic scenario, he said. In the meantime, he worries techies have sent the message that a class of humans will soon be obsolete. “People won’t say ‘You’re so kind,’ they’ll say, ‘I hate you, you’re telling me you’re needed and I’m not and I’m dependent upon your generosity.’”
Rhodes declined to comment on Altman’s world view and how it might have shaped the trial, noting that the study was not intended to be prescriptive. “There isn’t one ‘Here’s this one solution’ to any challenging problem,” she said. “There’s never one solution.”
Got a story tip for us? Contact Sarah Emerson on Signal at 510-473-8820 or email semerson@forbes.com.
But Altman is known for willing his visions into existence, sometimes at great cost, a characteristic that’s made him a polarizing figure in tech. In 2019, he founded Worldcoin, the iris-scanning cryptocurrency company he claimed would create a “collectively owned global currency that will be distributed fairly to as many people as possible.” The project has fallen far short of its goal to onboard 1 billion users by 2023, and has been mired by a parade of controversies. Now that he’s piloting the world’s most powerful AI company, it’s difficult to imagine even the most well-intentioned research project escaping his sphere of influence.
And then there are the actual entanglements. OpenResearch and OpenAI share DNA, as the AI company claims to have been supported by a donation from the lab. They have also occasionally shared staff — a former general counsel and one individual who simultaneously held different roles at each workplace. Last year, a researcher affiliated with both OpenResearch and OpenAI coauthored a study on AI’s effects on the labor market. OpenResearch said that because both organizations were founded at the same time, these were obvious opportunities for collaboration. Finally, the lab has just two board members: One of them is Altman; the other is Chris Clark, OpenAI’s former head of nonprofit and strategic initiatives. Clark left the AI company earlier this year, saying he wanted to “devote more time to the people and projects I care deeply about outside of OpenAI,” The Information reported in May. He remains OpenResearch’s COO, where he continues to manage high level operations across the organization.
OpenAI and Altman did not respond to a request for comment.
After completing a PhD in social work and political science at the University of Michigan, Rhodes responded to Altman’s 2016 job listing having “never heard of Sam or Y Combinator or anything.” He hired Rhodes that year, making her one of the first employees at YC Research, which would later become OpenResearch. The lab was established to incubate long term projects asking open-ended questions. It also housed a humanist technology hub led by computing pioneer Alan Kay and a “better cities” project. Most notably, it supported a team of artificial intelligence experts building OpenAI at their inception.
Since its launch in 2015, OpenResearch and its entities have amassed some $60 million in funding. Ten million came from OpenAI’s nonprofit arm, while Altman has donated $14 million through a $25 million line of credit with the lab, according to recent tax filings. Other supporters include Twitter cofounder and basic income advocate Jack Dorsey, who gave $15 million through his charitable foundation Start Small, and GitLab cofounder Sid Sijbrandij, who contributed $6.5 million. Through some of its researchers, the project has also received roughly $1.1 million in grant money from the National Institutes of Health and National Science Foundation.
The lab — which counts six full-time staff, six volunteer academic partners, and 17 volunteer advisory board members — has deliberately kept a low profile to avoid the shadow of its benefactors. But it attracted attention several years ago after launching two basic income pilots in Oakland, California, which were intended to surface issues that might arise during future, larger trials. Fewer than 100 people received up to $1,500 a month for roughly a year. In 2018, Wired called the effort slowgoing, and obtained an email Rhodes had sent to then-Oakland Mayor Libby Schaaf, saying, “Although it’s frustrating for funders, it has been good from a research standpoint.” OpenResearch declined to say who these frustrated funders were, but noted that the pilots were only meant to better their understanding of things like recruitment and transferring money to unbanked people.
Its latest 3,000-person trial ran from November 2020 through last October, and saw nearly 40,000 people respond to 1.1 million promotional mailers sent to addresses in Texas and Illinois. Applicants between ages 21 and 40 with household incomes not exceeding 300% of the federal poverty level were selected across urban, suburban and rural areas. (In his original callout, Altman said they were looking to enroll people “who are driven and talented but come from poor backgrounds.”) A third were randomly chosen to receive $1,000 a month, while the remaining control group received $50 a month.
One participant, Cara, was diagnosed with a rare nerve disorder that kept her from working. They had been on short term disability, had sold their possessions and had even set up a GoFundMe page to make ends meet. “It was like feeling the loss of being able to care for yourself,” they recalled in an interview recorded and shared by OpenResearch. Cara was assigned to the $1,000 trial group and said the monthly payments “took the panic a handful of notches down.”
The study’s broadest finding was that cash creates flexibility — to be more selective in job searches, to obtain medical care or to assist a family member with their own bills. Cash transfers led participants to spend more on basic needs like food, rent and transportation, results showed.
“What surprised me the most was that [compared to the $50 control group] the biggest increase in spending was on financial support to others,” said Karina Dotson, research and insights manager at OpenResearch. Dotson said participants reported using funds to give gifts and loans, donate to charity or to help incarcerated relatives. “And that was especially so for the lower income recipients in our population, who we know from existing literature are more likely to have low income social networks as well.”
The team also asked for permission to take blood draws, something 1,206 participants wound up doing. They measured biomarkers like cholesterol, diabetes risk and hypertension, but saw no significant changes. “Personally, I was not expecting to see in that short term an actual change in physical health,” said Rhodes. “Especially with this population that has perhaps had limited access to care for a long time.” What researchers did observe was a small increase in the likelihood that someone would seek healthcare, like a visit to the dentist’s office.
Other tech companies are getting involved with giving out free money; Google has funded a study on basic income and homelessness that will soon take place in the Bay Area. OpenResearch also contributed its expertise to help Illinois state representatives pass a law in 2019 that prevents participants of non-government funded cash transfer studies from losing their existing benefits.
Whether Altman plans to keep funding basic income research is unclear. OpenResearch has backed other projects that were eventually wound down, like a platform for promoting Covid-19 clinical trials. And it said it will continue to raise money to prolong its work on cash assistance and perhaps begin looking into medical inequality. Earlier this month, OpenAI announced a partnership with wellness company Thrive Global to build a personalized AI healthcare coach.
Meanwhile, Altman’s views on basic income seem to be evolving. A few months back, he floated a newer idea for humanity: financial dividends that “everybody” would receive from large language models like ChatGPT. He didn’t elaborate on how or why this would work, but has decided to call it “universal basic compute.”
