Biocomputing: Using Human Brain Cells for AI and Data Processing

Researchers in academia and the commercial sector are developing biocomputing systems that replace or augment silicon-based hardware with living biological matter. This field, which utilizes synthetic biology to create computer architecture, aims to address the increasing energy and data storage demands of modern artificial intelligence.

One approach involves the use of organoids, which are miniature clusters of lab-grown cells. These biological structures are integrated with electronic hardware to create what some researchers call wetware.

FinalSpark and the Neuroplatform

The Swiss company FinalSpark has launched a platform called Neuroplatform, which allows scientists to rent access to a computer powered by human-brain organoids for $500 a month. The company’s objective is to provide a sustainable alternative to current AI training methods, with co-founder Fred Jordan stating a goal of artificial intelligence for 100,000 times less energy than state-of-the-art generative AI currently requires.

The Neuroplatform architecture consists of processing units that each host four spherical brain organoids. Each organoid is approximately 0.5 millimeters wide and is connected to eight electrodes. These electrodes serve two functions: they electrically stimulate the neurons within the living sphere and link the organoids to conventional computer networks.

To mimic the natural reward system of the human brain, the neurons in these systems are selectively exposed to dopamine, a feel-good neurotransmitter. The biological components are derived from stem cells sourced from human skin cells, which FinalSpark purchases from an official supplier clinic in Japan.

Cortical Labs and the CL1

Another entity operating in the biocomputing space is Cortical Labs, which has developed the CL1, described as a code-deployable biological computer. Unlike the organoid-based approach of FinalSpark, Cortical Labs grows real neurons directly on custom silicon chips.

This integration of biology and traditional computing is designed to create systems that learn intuitively and require less energy and training data than conventional computers to master complex tasks. In 2021, Cortical Labs demonstrated this capability by growing neurons on a chip and training them to play the game Pong. The neurons were linked to a computer, allowing them to sense the position of a virtual ball and control a virtual paddle.

Technical Context and Industry Impact

Traditional AI depends on silicon-based hardware that has been the industry standard since the 1950s. Biocomputing seeks to move beyond this by combining real neurons with hardware to create processing systems that are more energy-efficient.

The potential applications for this technology extend beyond general computing. According to Cortical Labs, the energy efficiency of biocomputing could impact healthcare discoveries, including the development of enhanced cell therapies and personalized medicine.

Biocomputing combines real neurons with hardware, creating processing systems that use less energy and learn from smaller datasets than conventional computers.

Cortical Labs

As these systems evolve, the vision for the industry includes the possibility of data centers populated by living servers that replicate the learning processes of the human brain while utilizing only a fraction of the energy required by current silicon-based methods.