World models
2 writings
Mechanistic Indicators of Understanding in Large Language Models
Philosophical Studies. 2026. With Pierre Beckmann. doi:10.1007/s11098-026-02513-1
Draws on detailed technical evidence from research on mechanistic interpretability (MI) to argue that while LLMs differ profoundly from human cognition, they do more than tally up word co-occurrences: they form internal structures that are fruitfully compared to different forms of human understanding, such as conceptual, factual, and principled understanding. We synthesize MI’s most relevant findings to date while embedding them within an integrative theoretical framework for thinking about understanding in LLMs. As the phenomenon of “parallel mechanisms” shows, however, the differences between LLMs and human cognition are as philosophically fruitful to consider as the similarities.
Download PDF
World Modeling in Transformers
arXiv 2609.21748. 2026. With Pierre Beckmann and André Freitas. doi:10.48550/arXiv.2609.21748
Shows that behavioral failures can make a transformer appear to lack a world model even when it has learned faithful representations of its environment. In TaxiGPT, a transformer trained on random walks through Manhattan whose failures had been read as evidence of an incoherent internal map, mechanistic analysis and causal interventions reveal a model that represents intersections and streets, tracks its position, and navigates with a goal compass; its failures stem from interference between superposed intersection features, whose consequences are limited by an “affordance packing” that groups intersections with the same legal moves. Proposes mechanistic indicators for comparing models, which show world-modeling capacities emerging at different stages of training, and argues for a shift from asking whether a model has a world model to mechanistically studying its world modeling.
Download PDF