A Dedicated Team in Pursuit of Artificial General Intelligence

Who We Are

Beijing Institute for General Artificial Intelligence (BIGAI) is an innovative research & development non-profit organization affiliated with Peking University, Tsinghua University, and other leading academic institutions in China.

Our Mission​

Pursuing a unified theory of artificial intelligence to create general intelligent agents for lifting humanity.

Our Approach

We’re dedicated to developing AGI through the “small data, big tasks” approach, inspired by findings from diverse subjects such as cognitive science and developmental psychology.

Our Research & Development Areas

Computer Vision

Key research areas include deep scene understanding, general vision, embodied vision, active vision, visual cognition, and human-scene interaction.

Machine Learning​​

Key research areas include automated model learning, optimization, representation and hierarchical learning, and communicative learning.


Key research areas include multimodal perception, cognitive reasoning, task and motion planning algorithms, and hardware system development

Natural Language​

Key research areas include embodied language understanding, semantic and syntactic analysis, common-sense and dialogue reasoning, and dialogue strategy learning and generation.

Cognition and Reasoning​

Cognitive science research plays a vital role in realizing general intelligent agents, equipping them with human-like cognitive abilities and common-sense learning and reasoning skills.

Multi-agent System​

Key research areas include multi-agent learning, reinforcement learning, communication, cooperation, competition, and mental modeling to advance collective intelligence.

Simulated Autonomous Intelligence

Key research areas include autonomous intelligence, embodied reinforcement learning, cognitive computing, symmetric reality, computer graphics, multi-agent simulations, and human-computer interaction.

Tong OS

TongOS include perception module scheduling and joint interpretation, task planning, multimodal human-machine interaction, and adaptive communication technologies for human-machine-object integration networks.

Tong PL

TongPL aims to construct a Turing complete language to describe concepts and relationships among tasks, actions, values, logic, algorithms, and uncertainty along with corresponding runtime environments (compilers or interpreters).


Summer Palace
Summer Palace
Peking University
previous arrow
next arrow

Do you want to join us?

Please send us an email including your CV and other relevant materials. We will get back to you as soon as possible.

Scroll to Top