Human Neurons Trained to Play Doom, Showcasing Biocomputing Potential
- Cortical Labs trained a cluster of 200,000 human neurons to play the game Doom using the CL1 chip.
- The neurons learned to navigate and shoot in a simulated environment through electrical signals.
- Brett Kagan highlights the potential for biological systems to operate at approximately 20 watts, offering energy-efficient alternatives to traditional AI models.
- The lifespan of these neuron cultures is about six months, with future development subject to FDA and NIH regulations.
This experiment by Cortical Labs demonstrates the potential for biological computing systems that could complement existing AI technologies by significantly reducing power consumption. The use of human neurons in gaming applications like Doom provides insights into how these systems might be applied beyond entertainment, potentially impacting fields such as robotics and drug screening.
The ability of human neurons to learn complex tasks with low energy consumption suggests a promising direction for hybrid computing models that integrate biological components with conventional silicon-based technologies. (Source)
