A petri dish of human brain cells is currently playing Doom. Should we be worried?
Scientists create a 'dishbrain' that learns to navigate and shoot in the classic FPS without traditional silicon chips.
A team at Australian startup Cortical Labs has successfully trained a cluster of 200,000 living human brain cells to play the classic first-person shooter Doom. The system, often referred to as a 'dishbrain,' consists of neurons grown on a high-density microelectrode array—essentially a specialized glass chip. Unlike traditional AI, which runs on silicon processors, this biological computer uses the inherent electrical activity and learning capabilities of the brain cells themselves to process information and control the game. The cells receive simplified sensory input about the game world (like distance to walls or enemies) and learn through stimulation to output commands for movement and shooting, demonstrating a form of embodied intelligence in a petri dish.
This breakthrough is part of the growing field of organoid intelligence, which explores using biological neural networks for computation. The implications are significant for both neuroscience and technology. For researchers, it provides a novel model to study learning, memory, and information processing in a controlled, simplified brain environment. For the tech industry, it hints at a future of ultra-low-power biocomputers. While still in its infancy and far from replicating the complexity of a full brain, this work challenges the paradigm that intelligence and computation must be silicon-based, opening doors to hybrid systems that could one day outperform conventional AI in specific, adaptive tasks.
- The system uses 200,000 living human brain cells (neurons) grown on a microelectrode array chip.
- It learned to perform key video game actions like navigation, aiming, and shooting in Doom without a CPU or GPU.
- The research is a major proof-of-concept for using biological neural networks for goal-directed, interactive computation.
Why It Matters
This work pioneers organoid intelligence, potentially leading to energy-efficient biocomputers and new models for studying brain function and disease.