Shares

Why Do Neural Networks Become Modular?

Reflections on our latest Nature Machine Intelligence paper: moving from statistical associations to how learning demands shape network structure.

October 1, 2026

From Invisible Hand to the Undesignable Mind

Why the AGI era's defining shift is not stronger intelligence, but a new mechanism of social coordination through tokens, agents, and relational division of labor.

Two Paths to Intelligence

Distinguishing natural Type-I intelligence from abstract Type-II reasoning and their complementary learning approaches.

What's the next step for LLMs?

Exploring the limitations of current large models and the path toward autonomous scientific construction with AutoResearch.

What will we do next?

A series of questions and thoughts about the future of AI, neuroscience, and their intersection.

APR on Brain-inspired Learning

This report covers the latest research achievements in neurally inspired algorithms, models, and architectures.

Modularization for Brain and Brain-inspired Neural Networks

An invited talk at the Beijing Academy of Artificial Intelligence on modular structures in the brain.

Brain-inspired Intelligence: The Road to the Future

A talk at YikeTalks in Xiamen Science City on harmonizing AI with human intelligence.

The Future of Developing Our Brains

A TEDxChengdu talk on neurodevelopment and future brain technologies.

Control Theoretic Analysis on Brain Network

A guest lecture at Peking University on applying control theory to analyze brain dynamics.