Researchers at Cortical Labs, an innovative start-up in Australia, have embarked on an unprecedented journey by teaching human neurons grown on a chip to play the iconic video game Doom. This groundbreaking achievement represents a significant leap forward from their earlier milestone in 2021, where 800,000 neurons successfully played Pong. Now, with just 200,000 neurons, they have tackled a more complex gaming environment.
The Cortical Cloud Advantage
At the heart of this remarkable development lies the “Cortical Cloud,” an interface that enables developers to program the neurons using Python. This user-friendly API empowers independent researchers, like Sean Cole, to engage with biological computing without needing advanced degrees. In a mere week, Cole managed to teach the neurons to navigate the challenges of Doom, illustrating the accessibility of this technology.
Biological Efficiency vs. Silicon Speed
While modern silicon chips can execute trillions of operations per second, the biological “wetware” of human neurons possesses a distinctive advantage: efficiency. The human brain operates on a mere 20 watts of power, comparable to a dim light bulb. In contrast, a silicon supercomputer tasked with mimicking even a fraction of human cognitive functions would require an extensive power infrastructure.
Furthermore, silicon chips are rigid and deterministic, executing commands swiftly without consideration for outcomes. In contrast, biological neurons are inherently curious and adaptive, having evolved over billions of years to solve problems with minimal energy expenditure.
The CL-1 Biological Computer
Central to this innovation is the CL-1, a biological computer chip that integrates human neurons derived from induced pluripotent stem cells (iPSCs). These stem cells, typically reprogrammed from adult skin or blood samples, are differentiated in laboratories into functional cortical neurons. Once matured, approximately 200,000 neurons are placed onto a high-density microelectrode array (HD-MEA), forming a specialized silicon chip with thousands of interface points.
The CL-1 serves as a crucial bridge between biological cells and the external world. However, the neurons themselves perform the essential computations. Through synaptic plasticity, they adapt their connections over time to minimize chaotic feedback, effectively establishing a logic path aligned with the game’s objectives.
Programming Neurons: A New Approach
Programming the CL-1 system diverges significantly from traditional coding methodologies. Instead of providing a rigid set of instructions, the process involves supervised sensory feedback that leverages the neurons’ natural inclination to seek order. This democratization of biological programming enables individuals without extensive training to engage with complex biological systems.
Adaptive Learning in Real Time
Unlike silicon-based AI, which relies on massive training datasets, the neurons exhibit adaptive, real-time, goal-directed learning. They do not require extensive exposure to countless game instances to understand their environment. Instead, they learn through immediate interactions and feedback, responding to the game’s visual data, which is translated into patterns of electrical stimulation.
The chip sends electrical signals that represent the game state, and the neurons respond accordingly. If the neurons fire in a specific pattern, the character performs actions such as shooting or moving. This goal-directed learning mechanism rewards neurons for successful actions, effectively teaching them to thrive in a challenging digital landscape.
Natural Handling of Uncertainty
When humans perform tasks, our brains manage countless micro-adjustments in real-time. For example, if we lift a slippery glass, our brains instinctively adjust our grip. Traditional computers struggle with this level of nuance, relying on complex “if-then” statements. Biological computers, on the other hand, are naturally equipped to handle uncertainty and variability, making them more akin to human cognition.
Beyond Gaming: Future Applications
While the biological chip may not excel at playing Doom, its potential extends far beyond gaming. The ability to teach neurons to operate within a 3D game environment marks a crucial step toward employing biological computers for high-precision tasks, such as controlling robotic arms or managing intricate processes that challenge conventional silicon-based systems.
Dr. Brett Kagan, Chief Scientific Officer at Cortical Labs, emphasizes that these biological systems should not be misconstrued as “mini-brains.” Rather, they are sophisticated materials that process information in ways that silicon cannot replicate. As research progresses, the distinction between biological and artificial intelligence may blur.
Ethical Considerations
As biological computing advances, new ethical dilemmas arise. Many neurons used in these experiments originate from induced pluripotent stem cells, which retain the donor’s DNA. This raises questions about ownership and intellectual property. If these biological computers achieve commercial success, who possesses the rights to the genetic blueprint of the donors? Historical cases, such as that of Henrietta Lacks, serve as reminders of the importance of consent in scientific research.
The Future of Biological Computing
In the quest to create machines that mimic human behavior, we might have inadvertently succeeded by incorporating human elements within the computer itself. As the technology evolves, it opens the door to profound implications for artificial intelligence, consciousness, and the nature of life itself.
Key Takeaways
- Human neurons have been trained to play Doom, showcasing the potential of biological computing.
- The “Cortical Cloud” interface simplifies programming for researchers without extensive training.
- Biological computers excel in adaptive real-time learning, contrasting with traditional silicon AI.
- Ethical considerations surrounding the use of donor cells must be addressed in future developments.
- This innovation may lead to practical applications in robotics and complex task management.
In conclusion, the intersection of biology and technology heralds a new era of computing. As we explore the capabilities of living neurons, we must navigate the ethical waters that accompany these advancements. The future of biological computing promises not only innovation but also a reconsideration of what it means to be intelligent.
Read more → www.zmescience.com