Welcome
Artificial General Intelligence Lab.
Pioneering the Path to AGI Through Collaborative Multi-LLM Intelligence
About Us
The Artificial General Intelligence Lab advances the science of multi agent collaborative intelligence through the MACI framework, a systematic architecture that lifts large language models from pattern repositories (System 1) to reflective, reasoning systems (System 2) via orchestrated multi agent debate, memory, and validation.
Founded by Dr. Edward Y. Chang, ACM Fellow and IEEE Fellow, the lab has led multi LLM agent collaboration research since 2020. Dr. Chang has over two decades of leadership in large scale machine learning and healthcare AI, including serving as Director of Research at Google (2006–2012) and President of HTC DeepQ Healthcare (2012–2021). He has been an Adjunct Professor of Computer Science at Stanford University since 2019, where he teaches advanced AI and precision medicine, mentors the next generation of AI researchers, and leads work on consciousness modeling, ethical AI alignment, and persistent planning systems. He also begins service as co Editor in Chief of ACM Books in December 2025, shaping the publication agenda for the global computing research community.
The MACI framework rests on eight foundational pillars spanning consciousness theory, critical reasoning, emotion modeling, ethical adjudication, entropy governed dialogue, adaptive planning, transactional workflows, and polymathic knowledge integration. Together, these pillars provide a principled pathway toward trustworthy, interpretable artificial general intelligence.
Our book, Multi LLM Agent Collaborative Intelligence: The Path to Artificial General Intelligence (ACM Books, 2025), presents the theoretical foundations and practical implementations that demonstrate how orchestrated multi agent systems can achieve capabilities beyond individual models, including persistent memory, cross validation, reasoning coherence, and long horizon planning. It is the first pioneer AGI book built around a 17 chapter MACI roadmap, with comprehensive coverage of the concepts, architectures, and applications required to move from pattern driven LLMs to truly collaborative, system 2 style intelligence.