The cells are placed on a microelectrode array that analyzes neural activity and uses that information to move a virtual paddle. (Image credit: Cortical Labs via YouTube)
Researchers from biotech startup Cortical Labs have taught mini-brains grown from human brain cells how to play Pong using information translated from neural activity. The company is looking to harness the power of brain cells to fuel their biological computing endeavor to create neural networks that use the power of the brain instead of traditional AI. According to Coral Labs, "Biological neural networks can solve problems in unfamiliar situations – independent of acquired knowledge – due to their self-organizing properties."
To teach the mini-brains how to play the game, the researchers created a simplified Pong version with no opponent. The cells are placed on top of a microelectrode array that analyzes the neural activity, and a signal is sent to the right or left of the platform to indicate the ball's position. The neurons from the brain cell activity send that signal back to move the paddle. While this might not sound like a great accomplishment, consider, on average, it takes AI around 90 minutes to learn how to play Pong; it took the brain-dish 5 minutes.
Cortical Labs hopes to use their findings to develop sophisticated technology that combines living brain cells and silicon to create "fluid intelligence." This would allow them to design robust networks capable of resisting physical damage, scalable systems by growing additional neuron pathways and power-efficient processors that self-organize and restructure themselves to solve problems. It will be interesting to see how this technology evolves over the next few years.
Have a story tip? Message me at: http://twitter.com/Cabe_Atwell