https://www.nature.com/articles/d41586-025-02852-9
[AI Summary]: The La Caixa Foundation in Spain has implemented an AI-powered screening system for grant applications, using three algorithms trained on successful past proposals to identify applications with low chances of success. In 2025, the system flagged 122 out of 714 biomedical grant applications, with human reviewers rescuing 46 before the remaining 76 were rejected. While the foundation argues this reduces reviewer burden and improves efficiency, critics worry about potential breakdown in trust between researchers and funders, with concerns about AI-generated applications being evaluated by AI systems perpetuating training biases and eliminating genuinely novel ideas. The debate highlights broader tensions in academia as funders seek efficiency gains while researchers question the role of algorithms in crucial funding decisions.
- Author: David Adam
- Published: September 5, 2025
- Source: Nature News