If AI helps a worker produce more, what improves in that worker’s life? Higher output might lead to better pay or more control over the work. It could also lead to higher expectations with little benefit to the person doing it.
The new Institute for Human Flourishing proposes to study those outcomes. Launched during the United Nations General Assembly’s high-level week, the India-based research, action, and policy lab is supported by the Rockefeller Foundation and Schmidt Philanthropies and endorsed by the Government of India’s Expert Group on Human Flourishing.
Its plans include demonstrations of human-AI collaboration in workplaces, an observatory tracking worker experience, and public playbooks and policy recommendations drawn from the findings. A proposed Index of Human Flourishing would consider agency, dignity, capability, and economic security alongside income.
I welcome that approach. Much of the debate about AI and work depends on forecasts. Leaders also need evidence about the choices they can make now: which uses help people, what support those uses require, and who receives the benefits.
For a farmer, better access to market information is valuable if it improves a decision or strengthens their position when selling. An artisan might use AI for translation or marketing, but the result matters only if it helps the business in a way the owner can sustain. A micro-enterprise that becomes dependent on an expensive platform may face a new vulnerability even while gaining useful capabilities.
These are questions for the institute’s research, not outcomes its launch has established. The Rockefeller Foundation’s announcement argues that getting existing tools into wider use is a major constraint. The launch materials say nearly 90 percent of India’s workforce is informal and identify farmers, artisans, and micro-enterprises as groups that could benefit.
Reaching those workers will require understanding their actual conditions. A tool may perform well in a demonstration and prove impractical where connectivity is unreliable, training is limited, or the cost is too high. Research that stays close to those conditions could help organizations avoid investing in solutions people cannot use.
Employers also have choices to make about productivity gains. They can improve service, raise wages, lower prices, or retain more profit. The technology does not settle those decisions. Measuring adoption alone would tell us little about whether workers gained greater security or influence over their work.
The institute says its Livelihoods Lab will publish findings as an open public good regardless of outcome. Following through on that commitment would make the research useful beyond the programs that succeed. Organizations need to know when an intervention fails and why, especially before attempting it at a larger scale.
I remain optimistic about AI’s ability to expand what ordinary people can do. That optimism gives me a reason to take this research seriously. If we want better livelihoods, we should be willing to examine whether the tools and programs we support are producing them.
The institute has set a promising direction. Its contribution will depend on the evidence it produces and whether that evidence helps people make better decisions about how AI enters their working lives.