Computer powered by human brain cells learns to play DOOM
What's the story
A cluster of human brain cells has successfully learned to play the classic video game DOOM. The achievement, while not on par with human players, is a major step toward real-world applications of biological computers. These could potentially be used for tasks like controlling robotic arms. The breakthrough was achieved by Cortical Labs, an Australian company that had previously demonstrated the capabilities of neuron-powered computer chips by playing Pong in 2021.
Technological advancement
New interface allows easier programming of neuron-powered chips
Cortical Labs has now created an interface that simplifies programming these chips with Python, a popular programming language. An independent developer, Sean Cole, used this new tool to teach the chips how to play DOOM in just about a week. Brett Kagan from Cortical Labs expressed excitement over this development, noting its accessibility and flexibility compared to their earlier work with Pong which took years of scientific effort.
Performance analysis
Chips shouldn't be compared to human brains
The neuronal computer chip, which used about a quarter of neurons from Pong demonstration, performed better than a randomly firing player but not as well as top human players. However, Kagan noted it learned much faster than traditional silicon-based machine learning systems and could improve its performance with newer learning algorithms. He also clarified that while these chips are biological, they shouldn't be compared to human brains as they process information in unique ways that can't be replicated in silicon.
Technological leap
Experts weigh in on significance of breakthrough
Andrew Adamatzky from the University of the West of England, Bristol, UK, said that DOOM is way more complex than previous demonstrations. He added that successfully interacting with it shows real progress in controlling and training living neural systems. Steve Furber from the University of Manchester, UK also acknowledged this technological leap but noted that there are still many unanswered questions about how these neurons play games without eyes or prior knowledge.
Future potential
Potential real-world applications of biological computers
Yoshikatsu Hayashi from the University of Reading, UK, said this leap in capability brings us closer to real-world applications like controlling robotic arms with biological computers. He is already working on a similar computer made from jelly-like hydrogel. Adamatzky added that what makes this achievement exciting isn't just that a biological system can play DOOM, but also its ability to handle complexity, uncertainty, and real-time decision-making.