Not yet human-reviewed
This source record was written automatically by the research pipeline and has not yet been reviewed by the maintainer or integrated into a topic page. Treat it as a provisional summary and verify claims against the original source.
The Turing Transformation: Artificial Intelligence, Intelligence Augmentation, and Skill Premiums
Citation. Agrawal, A. K., Gans, J. S., & Goldfarb, A. (2023). The Turing Transformation: Artificial Intelligence, Intelligence Augmentation, and Skill Premiums (NBER Working Paper No. 31767). https://doi.org/10.3386/w31767
Summary. This NBER working paper develops a model in which automating a scarce, high-value task can expand access to complementary jobs for workers who previously lacked that task-specific skill. It calls this potential reduction in skill premiums and expansion of labor income a “Turing Transformation,” contrasting it with outcomes in which automation lowers wages or concentrates gains. The authors argue that distributional effects depend on which workers’ skills are automated or complemented, not on whether a technology is labeled automation or augmentation.
Key claims
- In the paper’s task-complementarity model, automating a scarce specialized task can permit more generic-skilled workers to participate in jobs previously restricted by that task and can raise total sectoral wages under its stated assumptions.
- The automation-versus-augmentation distinction does not by itself predict labor-market distributional effects; worker-skill heterogeneity and the allocation of complementarities determine who gains or loses.
- The paper uses transportation, call centers, medicine, translation, and writing to illustrate how AI that substitutes for one group’s task-specific skill may complement other workers’ broader job performance.
- The authors identify market power over AI and the empirical dominance of their proposed mechanism as open questions rather than settled outcomes.
Evidence & limitations
- The central evidence is a formal two-task model adapted from task-based economics plus illustrative cases and cited empirical studies; it does not estimate the net distributional effect of a particular AI deployment.
- This is an NBER working paper circulated for discussion and comment and has not been peer reviewed by the NBER Board of Directors.