August 8, 2024 By Jordan Kinard
In the search for less energy-hungry artificial intelligence, some scientists are exploring living computers
Artificial intelligence systems, even those as sophisticated as ChatGPT, depend on the same silicon-based hardware that has been the bedrock of computing since the 1950s. But what if computers could be molded from living biological matter? Some researchers in academia and the commercial sector, wary of AI’s ballooning demands for data storage and energy, are focusing on a growing field known as biocomputing. This approach uses synthetic biology, such as miniature clusters of lab-grown cells called organoids, to create computer architecture. Biocomputing pioneers include Swiss company FinalSpark, which earlier this year debuted its “Neuroplatform”—a computer platform powered by human-brain organoids—that scientists can rent over the Internet for $500 a month.
“As far as I know, we are the only ones in the world doing this” on a publicly rentable platform, says FinalSpark co-founder Fred Jordan. Initially bankrolled with funds from its co-founders’ previous start-up, FinalSpark seeks an environmentally sustainable way to support AI. “Our principal goal is artificial intelligence for 100,000 times less energy” than what’s currently required to train state-of-the-art generative AI, Jordan says. Neuroplatform uses a series of processing units hosting four spherical brain organoids each. Every 0.5-millimeter-wide organoid is connected to eight electrodes that electrically stimulate the neurons within the living sphere; those electrodes also link the organoids to conventional computer networks. The neurons are selectively exposed to the feel-good neurotransmitter dopamine to mimic the human brain’s natural reward system. These twin setups—positive dopamine rewards and electrical stimulation—train the organoids’ neurons, prompting them to form new pathways and connections much in the same way a living human brain appears to learn. If perfected, this training could eventually allow organoids to mimic silicon-based AI and serve as processing units with functions similar to today’s CPUs (central processing units) and GPUs (graphics processing units), FinalSpark says.
Research teams at 34 universities have asked to use FinalSpark’s biocomputers, and so far the company has provided access for scientists at the University of Michigan, the Free University of Berlin and seven other institutions. Each one’s project focuses on a different aspect of biocomputing. The University of Michigan team, for example, is investigating the electrical and chemical prompts necessary to change organoid activity—in effect creating the building blocks of an organoid-specific computer language. Scientists at Lancaster University Leipzig in Germany, meanwhile, are trying to fit the organoids into different models of AI learning.
Sticking points remain for organoid computing’s ability to compete with silicon on a large scale. For one thing, no standardized manufacturing system exists. And living brains die: FinalSpark’s organoids only survive for an average of around 100 days (and that’s considerable progress from the original experiment’s lifespan, which was just a few hours). But Jordan notes that Neuroplatform has “streamlined” its in-house process for making organoids, and its facility currently houses between 2,000 and 3,000 of them.
For now, the organoids and their behavior are live streamed 24 hours a day for researchers (and anyone else) to observe. “The challenge is to find the appropriate way to get neurons to do what we want them to do,” Jordan says…
…Adamatzky acknowledges that brain organoids “might offer advanced functionalities due to their complex and neuronlike structures” despite his advocacy of fungal computing. Jordan, for his part, is confident in FinalSpark’s choice for its biocomputers. Of all the cells to pick from, he says, “human neurons are the best at learning.”
Read the full article here.
Leave a Reply