Not long ago, Cortical Labs unveiled a video showcasing a cluster of human neurons in a Petri dish playing a simplified version of Pong. This innovative company has achieved a milestone by having cultured neurons engage with Doom.
The promotional video features “neuron gameplay,” illustrating Doom running on their CL-1 neural computing system, which hosts 200,000 human neurons arranged in a “multi-electrode array.” This comparison emphasizes the incredible complexity of our brains.
The video becomes increasingly fascinating as company representatives explain the intricacies involved. It’s neurons, not mere programming, controlling the game. Specific patterns of electrical stimuli are mapped to game data, with neurons sending signals to control character actions. For instance, a specific signal pattern triggers Doomguy to fire, while another prompts movement to the right.
A glimpse of the chip under the microscope reveals an intricate network of organic neurons woven around precise circuitry.
The neurons are in the process of learning, resembling novices similar to a player new to the game. However, as these neuron networks exhibit plasticity akin to human brains, they may become more adept at Doom over time.
Importantly, the neurons are operating Freedoom, a game built on the Doom engine, void of original game elements due to copyright constraints.
Such advancements raise ethical questions about the source of these cells, especially considering that human neurons contain DNA. We could face a future where neurocomputers are powered by the neurons of a single individual.
Cortical Labs has initiated the “Cortical Cloud,” a platform inviting developers to explore the capabilities of the CL-1 via a user-friendly API based on Python. This approach is admirable but raises concerns about unintended consequences, such as potential misuse of human neurons. The future of such technologies might feel as normalized as today’s sophisticated processors.