Can Living Human Brain Cells Power AI? | Bloomberg Primer

Key Concepts

Pong, AI, biocomputing, organic computing, biological hardware, wetware, DishBrain, brain organoids, neurons, silicon chips, Moore's Law, energy consumption, data centers, free energy principle, Neuroplatform, CL1 unit, neurodegenerative diseases, Parkinson's disease, Alzheimer's disease, ethics of biocomputing, consciousness.

The Promise and Challenges of Biocomputing

The video explores the emerging field of biocomputing, where living human brain cells are integrated with computing technology. This approach aims to leverage the efficiency and adaptability of biological systems to overcome the limitations of traditional silicon-based computing, particularly in terms of energy consumption and data requirements.

The Allure of Biological Efficiency

• Lower Energy Consumption: Human brains operate on approximately 20 watts, significantly less than the 40 megawatts required by supercomputers for similar tasks. This disparity highlights the potential for biocomputing to address the growing energy demands of AI and data centers. By 2034, data centers are projected to consume 1,580 terawatt hours annually, equivalent to the energy consumption of India.
• Reduced Data Requirements: Biological systems, shaped by evolution, can learn and adapt to new environments with far less training data compared to traditional machine learning algorithms.
• Overcoming Moore's Law: The video explains that Moore's Law, which predicted the doubling of transistors on an integrated circuit every two years, is approaching its physical limits as transistors reach atomic scales. Biocomputing offers a potential alternative to continue advancing computing power.

DishBrain: A Proof of Concept

• Cortical Labs' Experiment: Brett Kagan and his team at Cortical Labs created DishBrain, a system where 800,000 neurons grown on a silicon chip learned to play Pong.
• Mechanism: The chip translated the Pong game into electrical impulses that the neurons could sense. When the neurons missed the ball, they received unpredictable electrical stimuli, while successful hits resulted in predictable stimuli. This incentivized the neurons to improve their Pong-playing skills.
• Free Energy Principle: The success of DishBrain is attributed to the free energy principle, which suggests that biological systems strive to minimize surprise and predict their environment.

FinalSpark and the Neuroplatform

• Subscription-Based Access: FinalSpark, a Swiss biotech company, offers access to 16 brain organoids via a cloud computing network called the Neuroplatform.
• Real-Time Data Streaming: Researchers can remotely observe real-time data streamed from the neurons and send stimuli to the organoids.
• Applications: The Neuroplatform is used for research in robotics, teaching alternative computing, and other fields.
• Funding: FinalSpark is self-funded, contrasting with Cortical Labs, which has received significant investment from venture capital firms.

Cortical Labs' CL1 Unit

• Commercial Product: Cortical Labs is developing the CL1 unit, a processor containing human brain cells, available with neurons on a chip, organoids, or bio-engineered intelligence.
• Environmental Control: The CL1 unit provides the necessary environment for the cells, including feeding, waste disposal, and temperature regulation.
• Pricing and Launch: The CL1 unit is priced at around $35,000 and is expected to launch in March 2025.

Biocomputing for Biomedical Applications

• Drug Development and Disease Modeling: Thomas Hartung's lab uses brain organoids to model neurodegenerative diseases like Parkinson's and Alzheimer's.
• Replacing Animal Testing: Brain organoids offer a potential alternative to animal testing, which is often costly, time-consuming, and not always predictive of human responses.
• Understanding Neurodegenerative Diseases: Researchers aim to use brain organoids to understand the mechanisms of neurodegenerative diseases and develop new therapies. For example, they are exploring whether organoids affected by Alzheimer's forget learned tasks more quickly than healthy organoids.
• Economic Impact: Neurodegenerative diseases have a significant economic impact. The cost of Parkinson's disease in the U.S. alone is estimated at over $50 billion annually.

Ethical Considerations

• Consciousness and Suffering: The video raises ethical concerns about the potential for brain organoids to develop consciousness and experience suffering.
• Donor Consent: Questions arise about the validity of donor consent for stem cells used to create brain organoids, particularly if the technology evolves beyond what the donor could have imagined.
• Treatment of Sentient Organoids: Ethical dilemmas emerge regarding the treatment of potentially sentient organoids, such as whether it is acceptable to stop feeding them or kill them.

Challenges and Future Outlook

• Scaling and Engineering: Significant engineering challenges remain in scaling up biocomputing, including maintaining a stable environment for biological materials within a computing system.
• Investor Caution: Investors are generally cautious about deep tech, especially new fields like biocomputing, due to the high risk and long development timelines.
• Long-Term Perspective: The video emphasizes that the development of biocomputing will likely take decades, similar to the semiconductor industry.
• Hybrid Systems: The future of computing may involve hybrid systems that combine biological and artificial components.

Conclusion

Biocomputing holds immense promise for revolutionizing AI and addressing the limitations of traditional computing. While significant challenges remain, including ethical considerations and engineering hurdles, the potential benefits in terms of energy efficiency, adaptability, and biomedical applications make it a field worth pursuing. The integration of living brain cells with computing technology represents a paradigm shift with the potential to reshape our understanding of intelligence and disease.