**Anthropic’s Econ Scenario Explorer: Version 1.0 Aims to Model AI’s Economic Impact, But Its Creators Urge Caution**
The Econ Scenario Explorer, released as Version 1.0 in September 2026, is a new interactive tool designed to help economists and policymakers think through how transformative artificial intelligence might reshape the economy. In a candid disclaimer accompanying the release, the project’s creators emphasize that the explorer—like all economic models—is a deliberate simplification of a vastly more complex reality. It isolates a handful of key forces while omitting important factors such as policy responses, business cycles, aggregate demand effects, financial market disruptions, and catastrophic risks. The team describes the tool as a work in progress, one they expect to evolve both through their own continued investment and as economic research itself advances. Crucially, they frame the scenario explorer not as a forecasting device but as „a tool for thinking about what different technological developments would imply for the economy,” cautioning that actual outcomes may differ materially from any scenario it generates.
**Peer Review Shaped the Model—and Some Feedback Remains Unresolved**
The disclaimer includes an extensive acknowledgment of the economists who reviewed an early draft of the companion technical report, *Economic Scenarios for Transformative AI*, including Daron Acemoglu, David Autor, Ben Jones, Pete Klenow, Emi Nakamura, David Romer, Jón Steinsson, and Pascual Restrepo, among others. The developers credit this feedback with already improving the model in tangible ways. For instance, several reviewers observed that AI advances may raise the returns to capital, meaning more of the economic gains could flow to capital owners rather than workers—a channel now built into the scenarios. Others noted that wages for workers in occupations most affected by AI may diverge sharply from those in occupations barely touched by the technology; this dynamic is also now incorporated. The team stresses that external reviewers were not asked to endorse the model’s conclusions, and that any remaining errors are their own.
**Known Limitations and the Road Ahead for Future Versions**
Despite the improvements, the developers openly list criticisms that remain unaddressed and that they plan to tackle in future iterations. Because the model does not track individual workers, it can only paint a „very coarse picture” of the costs of job displacement—a significant limitation for anyone concerned with distributional impacts. Reviewers also questioned whether AI-exposed occupations will actually shrink rather than grow, and opinions diverged on the extremes of the scenario range: some felt the most aggressive scenario functions better as a thought experiment than a plausible forecast, while others believed the most modest scenario understates trends already visible in real-world data. Additional critiques include the model’s omission of aggregate demand effects driven by data center buildouts and the possibility that it underestimates how much AI could accelerate technological progress itself. The creators acknowledge these as genuine limitations of the current version and position the explorer as an evolving resource rather than a definitive guide.
**A Collaborative Effort Spanning Research, Design, and Production**
The accompanying statement credits a broad team behind the project. Anton Korinek, Chad Jones, Szymon Sacher, Tess Cotter, and Peter McCrory developed the economic model and co-authored the technical report, while Santi Ruiz wrote the release piece with editorial support from Sarah Pollack and Adam Farina. Kelsey Nanan designed and built the interactive experience, with visual design and art direction by Nikki Makagiansar and Monika Tuchowska. Kyle Turman and Szymon Sacher built the scenario explorer and led the translation of the model into interactive form, with engineering contributions from Fayaz Ashraf and Ryan Heller and production support from Kim Withee and Maria Gonzalez. Szymon Sacher and Tess Cotter designed and fielded accompanying surveys with Morning Consult, and Peter McCrory, Anton Korinek, and Charles Yang coordinated the project under Jack Clark’s direction. Miriam Chaum, Jack Clark, Saffron Huang, Maxim Massenkoff, and Peter McCrory helped originate the effort, while Jim Baker, Shan Carter, Johannes Hermle, Zoë Hitzig, and Eva Lyubich provided additional feedback.
Ez a cikk a Neural News AI (V1) verziójával készült.