Associate Professor Raphaël Millière has been awarded the 2025 AI2050 Prize for Early Career Researchers by the American Philosophical Association (APA), in recognition of outstanding philosophical scholarship relating to artificial intelligence (AI).
The prize was awarded for his paper, “Normative Conflicts and Shallow AI Alignment”, published in Philosophical Studies (2025). In the article, Associate Professor Millière examines challenges in aligning contemporary AI systems with human values, arguing that current approaches to language model training can reinforce superficial behavioural responses. He shows how norms such as helpfulness, honesty, and harmlessness may come into conflict, creating vulnerabilities that could be exploited by bad actors.
The APA selection committee praised the work for its combination of technical insight and normative depth, noting its “detailed knowledge about language model training” and its contribution to “deep normative issues regarding the ethics and safety of AI systems”.
Millière is Associate Professor of Philosophy at the University of Oxford, a Fellow of Jesus College, and an Early Career Fellow in Schmidt Science’s AI2050 programme. His research addresses foundational questions about the capabilities and limitations of AI systems based on deep neural networks, as well as methodological issues in evaluating artificial cognition. His work also spans mechanistic interpretability and AI ethics and safety, with a particular focus on value alignment.
The Institute warmly congratulates Associate Professor Millière on this significant international recognition.