What Is the Researcher's Hour Worth When AI Does Better? Comparative Advantage Put to the Test of a Dated Case
If an artificial intelligence (AI) outperforms a researcher in every task, what is left of comparative advantage? Two notes already published answered for a country, using Ricardo's four numbers. This article answers for a person, from an actual case: those two notes were written in two days by a researcher and an AI, and their entire conversation is dated. The division of tasks that was observed is the one the parable predicted. The AI did what it is far better at. The researcher kept what it is only slightly better at, and what the AI cannot produce: signing the text and answering for it. Comparative advantage therefore survives, but its role changes. It no longer says who does what best. It says how to use the only time that is scarce, the researcher's. Ricardo's question then becomes Becker's (1965): what is that hour worth? The article presents the measure proposed by Murphy and Topel (2006) for the time a person has left to live, and it raises, without settling it, the question of what AI changes in that measure. This record contains the English version and the French version of the article (« Que vaut l'heure du chercheur quand l'IA fait mieux que lui ? L'avantage comparatif à l'épreuve d'un cas daté »). The two notes it builds on are available at https://doi.org/10.5281/zenodo.24299462 and https://doi.org/10.5281/zenodo.24332433. Declaration of generative AI use: the author used Claude (Claude Fable 5.2, Anthropic) to reread the working conversation, to establish the dated table of decisions, to read the articles by Becker (1965), Becker, Philipson, and Soares (2005), and Murphy and Topel (2006), to propose the outline, and to draft the text, under his direction. The idea of a dated account was proposed by the AI, at the request of the author, who was looking for a new topic. The two questions that organize the article, that of comparative advantage and that of the value of the remaining time, as well as the recourse to Becker, come from the author. The author supplied the articles, approved the outline, reviewed and verified the content, and takes full responsibility for it.
Authors
- Fabrice Mazerolle
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-21
- DOI
- https://doi.org/10.5281/zenodo.24409675
- Primary Topic
- Labor market dynamics and wage inequality
- Type
- preprint