Yi Zeng

Yi Zeng (曾毅) stands at a fault line that most AI researchers prefer not to occupy: the seam where engineering specifications meet civilizational ethics. A cognitive scientist and AI safety scholar at the Institute of Automation of the Chinese Academy of Sciences, Zeng has spent his career insisting that the question of what we build must be settled before we decide how much of it to build. In 2019, he led the drafting of the Beijing AI Principles—fewer than two thousand words that introduced Confucian harmony as a first-order governance value alongside the Western liberal framework's individual rights. His BrainCog platform pursues spiking neural networks modeled on biological cognition, on the conviction that genuine intelligence cannot be separated from embodiment, self-perception, and social awareness. Where the dominant AI paradigm treats benevolence as a fine-tuning problem, Zeng treats it as an architectural constraint: the Confucian concept of ren—the immediate, prereflective responsiveness to the suffering of others—must be built into the system at the