Universal Basic Income and Labor Restructuring in the AI Era: From the Finnish Experiment to the Global Debate

In 2013, Carl Benedikt Frey and Michael A. Osborne of the University of Oxford published a paper that transformed global policy debates -- they estimated that 47% of U.S. jobs faced a high risk of being displaced by computerization.^[1] More than a decade later, the rise of generative AI has brought this prediction back into sharp focus. As AI moves beyond replacing physical labor and begins to encroach upon the domain of knowledge work, an age-old policy concept -- Universal Basic Income (UBI) -- is returning to the policy agenda with unprecedented force. This article takes no predetermined stance; instead, it uses the major UBI experiments completed around the world as an evidentiary foundation to analyze the feasibility and limitations of this policy in the AI era.

I. The Real Threat of Automation: 47% or 14%?

Frey and Osborne's study generated enormous resonance due to the sheer force of its conclusion -- 47% of American jobs fall within the "high-risk zone" of automation.^[1] However, subsequent OECD research employed a different methodology and arrived at significantly lower estimates. In their 2016 study, the OECD's Arntz, Gregory, and Zierahn argued that the unit of automation is not "occupations" but "tasks" -- most occupations contain only certain tasks that can be automated, rather than entire occupations being displaced.^[8] Using this task-level analytical approach, they revised the proportion of high-risk jobs across OECD countries to approximately 14%.

MIT economists Daron Acemoglu and Pascual Restrepo provided a more nuanced analytical framework.^[2] They pointed out that automation produces two opposing effects: the "displacement effect" eliminates the demand for human labor in existing tasks, while the "reinstatement effect" simultaneously creates new tasks and occupations. Historically, the latter has always exceeded the former -- agricultural mechanization eliminated vast numbers of farming jobs, but simultaneously gave rise to countless new occupations in industry, services, and the knowledge economy.

However, Acemoglu also sounded a warning: the critical issue is not aggregate quantity, but speed and distribution. If new tasks are created more slowly than old tasks disappear, or if the affected populations lack the skills and resources required for transition, then even if the long-term equilibrium is positive, the short-term social impact could be catastrophic. This is precisely the core premise of the UBI debate.

II. Global UBI Experiments: What Does the Empirical Evidence Say?

The Finnish Experiment (2017-2018): The Most Rigorous Randomized Controlled Trial

The Finnish government conducted the world's most rigorously designed UBI experiment from 2017 to 2018.^[3] Two thousand unemployed individuals aged 25-58 were randomly selected to receive a monthly basic income of 560 euros, with no employment conditions attached. The final evaluation report published in 2020 revealed three core findings:

1. Limited employment impact: The group receiving basic income worked approximately 6 more days during the first year of the experiment, but the statistical significance was limited. This refuted the concern that "UBI would drastically reduce the willingness to work," but it also failed to demonstrate that UBI could significantly promote employment.
2. Significant improvement in mental health: Participants' life satisfaction, mental health, and confidence in the future were significantly higher than in the control group. This is arguably the most consistent and compelling empirical finding regarding UBI.
3. Increased institutional trust: The group receiving UBI showed increased trust in social institutions and in other people, which carries potentially positive implications for social cohesion.

Kenya's GiveDirectly Experiment: Long-Term Evidence from a Developing Country

GiveDirectly's experiment in Kenya is one of the largest UBI experiments to date. Banerjee, Faye, Krueger, Niehaus, and Suri reported on the short-term results from 2017 to 2020.^[4] The study found that cash transfers significantly increased household consumption and asset accumulation, with beneficiaries investing a considerable proportion of the funds into productive activities such as small businesses. This evidence from an environment of extreme poverty demonstrates that the stereotype of "poor people will squander money" is inconsistent with empirical data.

The Stockton SEED Program in the United States (2019-2021): An Urban Experiment in a Developed Country

The SEED program in Stockton, California, provided 125 residents with $500 per month in unconditional cash transfers for 24 months. Preliminary analysis showed that full-time employment among recipients rose from 28% to 40% within one year (compared to a rise from only 32% to 37% in the control group); income volatility decreased; and mental health improved.^[9] Notably, these funds were primarily spent on food (37%) and everyday necessities (22%), rather than the luxury consumption that critics had feared.

III. Economic Arguments and Counterarguments for UBI

In his book Basic Income, Guy Standing presented the most systematic philosophical and economic case for UBI.^[5] He argued that under the dual pressures of globalization and automation, traditional employment-oriented social security has broken down -- an increasing number of people exist in a state of "precariat," holding jobs but with unstable incomes and incomplete benefits. UBI provides a universal safety floor.

The IMF's 2018 working paper offered a more cautious analysis.^[7] The study noted that UBI's core challenge lies in fiscal feasibility -- distributing an amount sufficient to sustain a basic standard of living to all citizens would require fiscal expenditures far exceeding existing social security budgets. The fiscal gap in most UBI proposals can only be bridged through substantial tax increases, cuts to other public spending, or a combination of both.

The World Bank's 2019 World Development Report further noted that the labor market changes wrought by new technologies require not only income protection, but also skills investment, lifelong learning systems, and a comprehensive overhaul of social protection institutions.^[6] UBI may provide a short-term buffer, but it cannot resolve the fundamental problem of skills mismatch.

IV. UBI in the AI Era: A New Logic of Argumentation

Generative AI has injected a new logic into the UBI debate. Traditional automation primarily affected routine, codifiable tasks (manufacturing, data entry, etc.), and the impacted populations were relatively concentrated and predictable. But generative AI has begun to encroach upon non-routine cognitive tasks -- writing, analysis, programming, legal research, design -- precisely the forms of knowledge work that were previously deemed "non-automatable."

This carries two important policy implications. First, the affected population is no longer limited to low-skill workers but now extends to mid- and high-skill knowledge workers, meaning traditional "upskilling" strategies may be insufficient. Second, the pace of transition may far exceed that of the past -- the Industrial Revolution took a century to complete its labor force transformation, whereas AI-driven transformation could unfold within a single generation.

In this context, UBI is being repositioned as a "transitional insurance mechanism" -- not a permanent replacement for the work ethic, but rather a means of providing individuals with the economic security to explore new directions, acquire new skills, and take entrepreneurial risks during a period of intense labor market reorganization.

V. Conclusion: UBI Is No Panacea, But It May Be an Essential Piece of the Puzzle

Synthesizing the empirical data from global experiments, the effects of UBI can be summarized as follows: it does not cause people to stop working (refuting the principal objection); it significantly improves mental health and quality of life (the most consistent positive finding); and fiscal feasibility remains the greatest challenge (especially at scale in developed countries).

In the AI era, UBI should not be regarded as a "silver bullet for the automation problem," but rather understood as one component within a broader reform of social protection -- complemented by skills retraining, lifelong learning systems, progressive automation taxes, and the modernization of labor market institutions, together forming a resilience framework for confronting technological change. As the World Bank emphasized, the core of a new social contract is not to protect old jobs, but to protect the people who do the work.^[6]