For a company often accused of moving too fast to measure its own effects, this is a notable step. On June 9, OpenAI launched the Economic Research Exchange, a platform that hands outside academics structured access to study how AI is reshaping work, productivity, and the broader economy. Selected researchers will run project-based collaborations with OpenAI's economics team, working with privacy-protected data and publishing what they find as independent evidence.
Applications are open until July 5, with selections announced by July 31. OpenAI says proposals will be judged on methodological rigor, feasibility, and whether they can produce credible evidence on how AI affects workers, firms, and institutions.
A companion to a bigger commitment
The exchange does not stand alone. The OpenAI Foundation has committed $250 million to fund research on how AI is changing jobs and communities, and to help people adjust to those changes. Most of that money is earmarked for studying which kinds of work are exposed to automation, with writing, analysis, customer service, human resources, and scheduling named among the categories under the microscope.
Together the two efforts read as an attempt to build a credible evidence base rather than to settle the argument by assertion. That is a meaningful distinction. The loudest claims about AI and jobs, in both directions, tend to outrun the data. A structured research program with independent publication is how you close that gap, assuming the independence holds when the findings are unflattering.
The obvious tension
It is fair to ask whether a company can fund research into its own labor-market footprint and stay neutral. OpenAI has an interest in the story landing a certain way, and the design of the exchange, with the company selecting projects and supplying the data, leaves room for skepticism. The safeguard is publication. If outside economists can publish results that cut against OpenAI's preferred narrative, the program earns its credibility. If they cannot, it becomes marketing.
The early-labor-market signals are still murky. Recent work from the New York Fed found only faint traces of AI in job-posting data, a useful reminder that the disruption everyone talks about has not yet shown up cleanly in the numbers. That uncertainty is exactly the gap programs like this are meant to fill. It also relates to a fear we have examined before, the claim that AI has already replaced programmers, which the evidence does not support.
Commentarii · 0