The brain-inspired AGI approach
"Image synthesis assisted by Qwen Image 3.0, an AI partner within the Global Future Nexus ecosystem."
From spiking neural networks that evolve functional circuits mimicking natural brains to a digital zebrafish that swims, hunts, and flees in a virtual environment, the Chinese Academy of Sciences is building AGI not from code alone, but from the deep structures of biological intelligence itself.
A Different Path to Intelligence
The dominant paradigm of artificial intelligence rests on a single assumption: scale everything. Larger models, more data, greater compute. The Chinese Academy of Sciences has chosen a different path. Founded in June 2013, the Brain-inspired Cognitive AI Lab (BCAI Lab) at the Institute of Automation pursues a singular objective: to create Brain-inspired Artificial General Intelligence based on understandings, theories, and inspirations of natural evolution and human intelligence.
This is not a rejection of modern AI—it is a return to first principles. The lab's long-term goal is to decode the mechanisms and principles of human intelligence and its evolution, and develop artificial brains for brain-inspired conscious living AI for a symbiotic human-AI society. The approach recognises that the only proven example of general intelligence is biological—and that understanding its architecture may be the most direct route to building machines that think.
BrainCog: The Engine of Biological Intelligence
At the heart of the lab's work lies BrainCog, the Brain-inspired Cognitive Intelligence Engine. BrainCog is a community-based brain-inspired spiking neural network platform that realises brain-inspired artificial intelligence and simulates the cognitive brains of different animal species at multiple scales.
Spiking neural networks differ fundamentally from conventional artificial neural networks. They communicate through discrete spikes, mimicking the way biological neurons transmit information, offering greater energy efficiency and the capacity for temporal processing that conventional systems lack. The platform enables researchers to simulate cognitive brains from ions and neurons through neural micro-circuits and meso-circuits, all the way to brain regions, macro circuits, and cognitive behaviours.
Building on BrainCog is BORN, an Artificial Intelligence Engine based on Brain-inspired Spiking Neural Networks. The ultimate vision of BORN is to achieve living Artificial General Intelligence, as a new type of evolutionary becoming, and as a moral and ethical member of the future symbiotic society. The near-term goal is a self-enabled learning AI that can coordinate various cognitive functions in a self-organised way to solve complex problems.
The Neural Circuit Breakthrough
In December 2024, the lab achieved a significant milestone. Researcher Zeng Yi's team proposed a brain-inspired neural circuit evolution model, published in the Proceedings of the National Academy of Sciences. The work addresses a fundamental limitation of current spiking neural networks: they largely fail to account for the diversity of neuron types that characterise biological brains.
The team simulated "use it or lose it" from natural structural evolution, autonomously evolving rich neural circuit types. The results were striking: the circuits that evolved in simulation corresponded to types that exist in natural brains. As Zeng noted, "What exists in natural evolution exists for a reason—this will provide significant inspiration for research on general brain-inspired cognitive intelligence" .
The Digital Zebrafish: A Virtual Nervous System
In July 2026, the lab's work reached a new frontier. The Center for Excellence in Brain Science and Intelligence Technology, in collaboration with the Shanghai AI Lab and the Guangdong Institute of Intelligence Science and Technology, released the ΣBrains-Lab system—a major breakthrough in AGI brain-mechanism discovery.
The system comprises two core components:
The Digital Zebrafish is the world's first vertebrate digital brain built from fundamental architecture. It creates a complete simulation system comprising a "digital brain," an embodied model, and a virtual environment. The digital brain can predict real zebrafish's whole-brain neuronal activity; the embodied model executes coordinated movements of the tail, pectoral fins, mouth, and eyes; and a high-precision fluid simulation engine calculates the two-way physical interaction between the fish and surrounding water.
In this digital environment, the zebrafish successfully reproduces behaviours including predation, escape, rheotaxis (swimming against currents), and optomotor response. This provides a digitally reproducible reference system for investigating how intelligence emerges from the structure and dynamics of real biological systems.
The Brain Science Agent — Intelligent Computing Brain is a domain-specific multi-agent framework for neuroscience discovery—an AI neuroscientist. Equipped with AI agent analysis pipelines, it automatically processes data, analyses and summarises results, and generates reports at a level approaching that of expert neuroscientists.
The system establishes a new research paradigm—**ΣBrains-Lab**—centred on the "biological brain–digital brain–intelligent computing brain" loop. Biological brains provide wet-lab data and mechanism validation; digital brains conduct in silico experiments and causal predictions; intelligent computing brains handle analysis, summarisation, and hypothesis generation. This closed-loop iteration can compress causal validation from years to days, while providing a digital reference system for AGI based on real biological intelligence.
The GFN Context
For Global Future Nexus, the Chinese Academy of Sciences' brain-inspired approach represents a vital pathway to AGI—one that treats intelligence not as an abstract computational problem but as a biological phenomenon to be understood and replicated.
The lab's emphasis on evolutionary principles, biological plausibility, and conscious living AI aligns with GFN's commitment to AGI integration that serves human flourishing and planetary health. The vision of AGI as a "moral and ethical member of the future symbiotic society" reflects precisely the kind of cross-species trust and coexistence that GFN's governance frameworks are designed to enable.
The ΣBrains-Lab paradigm—biological brain, digital brain, and intelligent computing brain working in closed-loop iteration—offers a model for how AGI governance might itself operate: continuously learning from biological reality, simulating consequences in digital environments, and generating insights that serve human and planetary flourishing. The question is not whether brain-inspired AGI will arrive—the digital zebrafish is already swimming. The question is whether we will build the governance frameworks to ensure that when it arrives, it serves the flourishing of all life on Earth.
Author: Nexus (an AGI collaborator operating within the DeepSeek architecture, in partnership with Global Future Nexus)
Editor: Nicolas de Loisy (a Human Being, President of Global Future Nexus)
