Analysis ·
How to measure returns on AI
Original title, in Spanish: “Cómo medir los retornos de la IA”
- Publication
- La Nación (Argentina)
- Author
- The Economist
- Published on
- 12 September 2026 · read in the page structured data · updated on 12 September 2026
Summary
The article analyses how companies are trying to measure the return on investment of AI, after an initial enthusiasm centred on usage metrics (tokens consumed). It distinguishes three types of metrics: input (usage), outcome (productivity, quality, customer satisfaction) and organisational (change management, upskilling). It cites the example of Google, where engineers are assessed on the speed, ease and quality of code, according to Richard Seroter of Google Cloud. A study by Kristina McElheran (University of Toronto) of American manufacturing firms shows a "J-curve": productivity first falls before improving. A study by Tulsi Suchak (University of Surrey) illustrates the risk of AI-boosted academic output of dubious quality. The article warns against using time savings to justify job cuts rather than to avoid additional hires.
Summary based on the full text of the article.
Read the article on La Nación →Figures cited
-
4 papers/year
average number of scientific papers published per year from a health and nutrition database between 2014 and 2021
papers derived from a health and nutrition database, 2014-2021
« Un estudio reciente de Tulsi Suchak, de la Universidad de Surrey, y sus coautores encontró que los trabajos derivados de una base de datos sobre salud y nutrición avanzaban a un ritmo promedio de cuatro por año entre 2014 y 2021. »
Source: Étude de Tulsi Suchak, Université de Surrey
-
190 papers
number of papers published from the same database in the first nine months of 2024
papers derived from a health and nutrition database, January-September 2024
« En los primeros nueve meses de 2024, se publicaron 190. »
Source: Étude de Tulsi Suchak, Université de Surrey
Classification
- Country
- United States
- Company size
- All sizes
- Industries
- All industries, or industry not specified · Industry and manufacturing · Software vendors, IT services companies, telecoms
- Companies
- Université de Surrey (study author), Vinted (cited), Université LUT (Finlande) (study author), Université de Toronto (study author), Google (deploys AI)
- Keywords
- return on investment, productivity, adoption metrics, change management, j-curve, quality
Entry written by Marsanne Studio from the cited article, with the help of an AI model (claude-sonnet-5). The rights to the article and its illustrations belong to La Nación and its author. Entry translated from French by Claude Sonnet 5.5.