Descriptions:
The AI Search channel explores Cortical Labs’ CL1 chip — a biological computing system powered by approximately 200,000 living human neurons that researchers have trained to play the video game Doom. The video traces the development lineage from the earlier Dishbrain prototype (2021–2022), which required up to 1 million neurons and took 18 months to show reliable learning behavior, to the CL1, which achieves stable operation with 80% fewer neurons while running more complex tasks.
Central to the CL1’s viability is a microfluidic perfusion circuit — a miniaturized automated plumbing system that continuously delivers glucose and nutrients to the neurons while removing metabolic waste, effectively functioning as a synthetic circulatory system. The neurons grow flat across a high-density multielectrode array (HDMEA) that enables two-way electrical communication: stimulating specific regions to send inputs into the living network and recording natural neuron firing at submillisecond voltage resolution to read outputs. This interface is what turns a dish of cells into an interactive computational system.
The broader argument is that biological neurons are inherently more energy-efficient and adaptable than silicon GPUs, potentially addressing the hard physical limits that transistor miniaturization is approaching. While CL1 remains firmly in early research territory, Cortical Labs’ demonstration of a functional living neural chip capable of interfacing with and learning from digital systems represents a notable step toward neuromorphic computing as a practical alternative to conventional silicon architectures.
📺 Source: AI Search · Published March 13, 2026
🏷️ Format: Deep Dive
