The World Inside Neural Networks

How neural geometry will unlock understanding and control of AI

The World Inside Neural Networks Research The Neural Geometry Series → The World Inside Neural Networks How neural geometry will unlock understanding and control of AI Authors Atticus Geiger * Ekdeep Singh Lubana * Thomas Fel * Jack Merullo * Michael Jae Byun * Owen Lewis * Tom McGrath * Correspondence to atticus@goodfire.ai, ekdeep@goodfire.ai Published May 7, 2026 --> Our world is full of objects and richly structured geometric relationships—the color wheel, road networks, family trees, months of the year—and humans use structured mental concepts to represent and navigate this complex realit

saved by

• Uzay Girit
• Lydia Nottingham
• Timothy Kostolansky
• Vihan Tiwari
• Rohan Prakash
• Jack Hogan
• Sean Lee

related reading

• Steering Along Manifolds to Control Neural Networksgoodfire.ai
• The Neural Geometry Series - Goodfiregoodfire.com
• Verbalizable Representations Form a Global Workspace in Language Modelstransformer-circuits.pub
• Toy Models of Superpositiontransformer-circuits.pub
• Neural Networks, Manifolds, and Topology -- colah's blogcolah.github.io
• Zoom In: An Introduction to Circuitsdistill.pub
• A Gentle Introduction to Graph Neural Networksdistill.pub
• Uncovering Neural Geometry in Vision Models With Block-Sparse Featurizersgoodfire.ai
• The Building Blocks of Interpretabilitydistill.pub
• When Models Manipulate Manifolds: The Geometry of a Counting Tasktransformer-circuits.pub
• What Is ChatGPT Doing … and Why Does It Work?-Stephen Wolfram Writingswritings.stephenwolfram.com
• [2210.13382] Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Taskarxiv.org