Scientists are exploring the field of biocomputing, which uses living human neurones to power computers, in a ground-breaking step towards more sustainable AI. This innovative approach, led by Swiss company FinalSpark, aims to reduce the massive energy demands of traditional silicon-based AI.
FinalSpark has introduced the “Neuroplatform,” a pioneering biocomputing system that uses lab-grown brain organoids—miniature clusters of human brain cells—as its processing units. For $500 a month, researchers can rent access to this cutting-edge technology, which represents a significant leap towards eco-friendly AI solutions.
The Neuroplatform operates with a series of brain organoids, each connected to electrodes that stimulate the neurons with electrical impulses and dopamine, mimicking the brain's natural reward systems. This setup allows the organoids to form new neural connections, potentially enabling them to perform tasks similar to current CPUs and GPUs but with vastly reduced energy consumption.
Despite its promise, biocomputing faces challenges. The organoids, which currently have a lifespan of around 100 days, lack standardised manufacturing processes and present ethical concerns about consciousness and the use of human brain cells for non-medical purposes. Nevertheless, FinalSpark’s facility manages to maintain between 2,000 and 3,000 active organoids at any given time.
Research teams from 34 universities are exploring various aspects of biocomputing using FinalSpark’s platform. Projects range from developing organoid-specific computer languages to integrating organoids into AI learning models. Meanwhile, researchers like Ángel Goñi-Moreno and Andrew Adamatzky are investigating alternative biological systems, such as bacterial and fungal computing, which might offer unique advantages and further diversify the field.
As the debate over the ethical implications of using human neurons continues, FinalSpark remains confident in the potential of its biocomputing approach.
The goal is to achieve an AI system that requires 100,000 times less energy than current methods, potentially revolutionising the way we approach artificial intelligence and sustainability.
