To date, AI’s energy usage is one of the most concerning aspects of the technology, with some predictions putting it’s nupcoming global energy needs in the double digits. There’s already been a lot of efficiency improvements… But this piece of news was somewhat unexpected!
In the ever-evolving world of artificial intelligence, a new frontier is emerging: biocomputing. According to a recent article by Jordan Kinard published on Scientific American, researchers are exploring the use of living biological matter, specifically human neurons, to create computer architecture. This innovative approach aims to address the growing energy demands of AI systems.
How It Works
Swiss company FinalSpark has introduced a “Neuroplatform,” a computer platform powered by human-brain organoids. These organoids are clusters of lab-grown cells that mimic the human brain’s learning processes. Each organoid, about 0.5 millimeters wide, is connected to eight electrodes that stimulate the neurons and link them to conventional computer networks. By exposing these neurons to dopamine, researchers can train them to form new pathways, similar to how a human brain learns.
Benefits
One of the primary benefits of this technology is its potential to drastically reduce the energy required for AI training. FinalSpark aims for their biocomputers to use 100,000 times less energy than current silicon-based systems. This could make AI more environmentally sustainable. Additionally, the Neuroplatform is available for rent at $500 a month, making it accessible for research teams across the globe.
Concerns
However, there are several challenges and ethical concerns. For one, the organoids have a limited lifespan, surviving only around 100 days. There’s also the ongoing bioethical debate about the potential for these mini-brains to gain consciousness. FinalSpark is actively seeking philosophers and researchers to address these ethical questions.
Possible Business Use Cases
- Environmental Monitoring: Develop biocomputers that can be submerged in water bodies to monitor environmental conditions in real-time.
- Healthcare Diagnostics: Create diagnostic tools that use organoid-based computing to analyze complex medical data more efficiently.
- AI Training Services: Offer a subscription-based service for companies to train their AI models using energy-efficient biocomputers.
As we stand on the brink of this new technological era, it raises an intriguing question: Could biocomputing be the key to making AI both more powerful and more sustainable?
Image Credit: FinalSpark (Four clusters of living neurons are connected to electrodes on FinalSpark’s Neuroplatform chip.)
