Anthropic just laid out where it wants the hard economic questions about AI to go next. The company published a research agenda for its Economic Futures Research Fund, according to Anthropic, setting the priorities and funding direction for outside researchers studying how AI reshapes work, wages, and growth. This is Anthropic putting money and structure behind a question most labs prefer to hand-wave: what happens to the economy when the technology actually lands.
What stands out here is that a frontier AI company is funding independent study of its own product’s fallout. That’s rare. Most of the AI economics conversation so far has run on vibes, viral charts, and dueling predictions. Anthropic is trying to replace some of that noise with empirical work.
What the agenda covers
Anthropic frames the research around a few core threads:
- Labor markets and jobs. Which tasks AI takes over, which it complements, and how that shows up in real hiring and wage data rather than forecasts.
- Productivity and growth. Whether AI actually moves the big economic numbers, and how fast those effects spread across industries.
- Policy responses. What governments, educators, and businesses can do to smooth the transition instead of reacting after the damage is done.
The fund backs external researchers, so the findings come from economists and academics rather than Anthropic’s own marketing team. That independence is the whole point.
Why this matters now
The status quo has been a mess of confident guesses. One camp says AI wipes out entire job categories by next year. Another says it’s overhipped and nothing structural changes. Neither side has much hard data, because the tools are new and the measurements barely exist. Anthropic’s agenda is a bet that we can do better than guessing if someone funds the measurement.
This is significant because policy is already being written on top of that thin evidence. Lawmakers are drafting AI rules, companies are restructuring around automation, and workers are making career decisions, all with weak data underneath. Better research doesn’t just satisfy academics. It changes what gets built into law and corporate strategy.
It also signals where Anthropic sees the real risk. The company has always leaned hard on safety framing. Extending that to economic disruption says the leadership treats labor-market shock as a first-order problem, not a footnote.
The Future Cast: what the next few years look like
Here’s where this points over the next one to three years.
Expect the AI-and-jobs debate to get more evidence-based and less theatrical. As funded studies start landing, the wild predictions on both ends will run into actual numbers. That’s healthy. It also means the findings could get uncomfortable, for Anthropic and everyone else, if the data shows sharper displacement than the optimistic story allows.
Expect policymakers to lean on this kind of work. When governments look for a factual basis to regulate AI or design retraining programs, independent research funded through efforts like this becomes the reference material. Whoever produces the credible data shapes the rules.
And expect other labs to feel pressure to match the move. Once one frontier company funds open study of its economic impact, staying silent starts to look like avoidance.
What to do about it
A few practical takes:
- If you build with AI: watch this research as it publishes. Real task-level data on what AI automates well beats guessing which workflows to reshape.
- If you lead a team: the smart planning question isn’t “will AI replace people” but “which specific tasks shift, and how fast.” That’s exactly what this agenda aims to measure.
- If you’re thinking about your own career: the complement-versus-replace split is the one to track. The jobs that pair with AI look very different from the ones it swallows.
Anthropic is trying to turn a shouting match into a research field. Whether the findings comfort the optimists or the doomers, they’ll be a lot harder to dismiss than the current round of predictions. Full details on the agenda are available at the original source.