Human brain cells are trouncing computers in raw speed and power
>
Scientists are using human brain cells to create biocomputers that offer “unprecedented advances in computational speed, processing power, data efficiency and storage capabilities” over the current zeitgeist in computer processing, artificial intelligence (AI).
By an article (opens in new tab) in the academic journal Frontiers in Science (FiS), scientists note that this form of biocomputing, known as organoid intelligence (OI), is a natural progression because AI was inspired by our understanding of the human brain.
OI is powered by lab-grown cell cultures known as brain organoids, three-dimensional clusters of brain cells, dividing structures, such as neurons and other cells that power our memory and learning abilities.
Organoid intelligence
This certainly sounds impressive, but like AI, OI itself needs to break through the skepticism barrier. After all, isn’t the reason why we put off most tasks in our daily lives on computers precisely because they work faster than we do?
Thomas Hartung, a professor at Johns Hopkins University’s Bloomberg School of Public Health, thinks making computers think like humans is less effective than simply putting the human element at the center of a computer.
“Silicon-based computers are definitely better with numbers,” explains Hartung. “For example AlphaGo [an AI that beat the world’s #1 Go player in 2017] was trained on data from 160,000 games. Someone would have to play five hours a day for over 175 years to experience that many games.”
“[But] we’re reaching the physical limits of silicon computers because we can’t fit more transistors into a small chip. The brain is wired completely differently. It has about 100 billion neurons connected through more than 1015 connection points. It is a huge power difference compared to our current technology.”
Hartung also claimed that brains are more energy efficient than the computers that power AI models. “For example, the amount of energy expended training AlphaGo is more than it takes to sustain an active adult for 10 years.”
OI is still in its infancy and has several shortcomings, not the least of which is that brain organoids currently contain about 50,000 cells, when, to be practical, this should be scaled up to “10 million,” Hartung said.
And as with AI, there are ethical concerns, although the idea that real human brains could evolve consciousness is perhaps more pressing than the idea that a computer could.
Still, one of the co-authors of the FiS paper, Dr. Brett Kagan, in December 2022 a study (opens in new tab) where a flat brain cell structure taught Pong to play, and the scientific community has bigger plans for the technology.
For example, Hartung notes that “personalized brain organoids” can be grown from adult skin cells, allowing scientists to study the effects of neurological disorders, such as Alzheimer’s disease, and test the effects of certain substances on learning and memory processing.
The challenge from here, he says, is building a scientific community willing to explore OI further.
AI having its day in the sun suggests this is a realistic goal, but in all likelihood we still have a long time – years, maybe decades – to go before the technology is in any way usable or tolerable within an enterprise environment. That’s never stopped us from reporting on cool, far-future developments like DNA storage sooner.