Human Brain Cells on a Chip Learn to Play Doom in Under a Week
Key Takeaways
- ▸Cortical Labs' human brain cell chips learned to play Doom in about a week using a new Python programming interface, compared to years of work required for their 2021 Pong demonstration
- ▸The neuronal computer used approximately 800,000 living brain cells and learned faster than traditional silicon-based AI systems, though performance remained below human level
- ▸The breakthrough demonstrates biological computers can handle complex, real-time decision-making and brings practical applications like robotic arm control significantly closer to reality
Summary
Australian company Cortical Labs has achieved a significant breakthrough in biological computing by successfully programming human brain cells on a chip to play the classic first-person shooter game Doom. The achievement, accomplished by independent developer Sean Cole using a new Python interface in approximately one week, represents a dramatic leap forward from the company's 2021 demonstration where similar neuronal chips learned to play Pong after years of painstaking effort.
The biological computer chip used roughly 800,000 living brain cells grown on microelectrode arrays capable of sending and receiving electrical signals. While the neurons' performance didn't match human players, they played better than random chance and learned significantly faster than traditional silicon-based machine learning systems. Brett Kagan of Cortical Labs emphasized that the true breakthrough lies in the accessibility and flexibility of the new programming interface, allowing developers with limited biological expertise to work with these systems.
Experts in the field view this as a substantial step toward practical applications of biological computing. The increased complexity of Doom compared to Pong demonstrates that living neural systems can handle real-time decision-making, uncertainty, and complex environments. Researchers suggest this brings biological computers closer to useful real-world applications such as controlling robotic arms, with the game-playing task serving as a simplified version of the spatial reasoning and motor control required for such applications.
- The new programming accessibility allows developers without extensive biological expertise to work with neuronal computing systems, dramatically democratizing the technology
Editorial Opinion
This development represents a pivotal moment in biological computing, not just for the technical achievement but for its implications about the convergence of living systems and computation. While the performance gap between these neuronal chips and conventional AI remains significant, the speed of learning and energy efficiency advantages suggest biological substrates may offer unique computational properties that silicon cannot replicate. The ethical questions surrounding the use of human neurons for computing tasks will only intensify as these systems become more capable and widespread.
