As companies around the world race to build more data centres to power artificial intelligence (AI) models, researchers are exploring whether living human cells could be used in computing systems.
An Australian start-up says it has created the world’s first device that allows users to “run code” on living human brain cells.
Cortical Labs has developed a system that combines lab-grown neurons with silicon hardware, allowing users to explore applications ranging from neuroscience and disease modelling to robotics and artificial intelligence (AI).
The system, CL1, works by growing neurons from stem cells and placing them on chips that can send and receive electrical signals.
“We're using these cells more like an engineering approach to build something that's never really existed before and might have properties that we've never been able to use before. And so far, the results are very exciting,” Brett J. Kagan, the chief scientific officer and chief operating officer at Cortical Labs, Euronews Next.
“All you need is a little bit of blood or some skin, and you can generate an indefinite supply of these cells that you can then turn into neurons,” Kagan added.
The company says it is working on biological computing facilities in Melbourne and Singapore, where multiple units of its system could be deployed and accessed remotely.
How is it different from the conventional silicon chip?
CL1 allows users to interact directly with the neurons, sending electrical signals as inputs and interpreting how the cells respond in real time.
Similar to conventional computing systems, it uses silicon chips, but they are equipped with microelectrodes that communicate with living neurons, sending signals and reading their responses as part of the computation.
Unlike conventional silicon-based computers, the shoebox-sized system uses living cell cultures that require a nutrient-rich liquid to survive, an approach sometimes described as “wetware”.
Some 120 units of such a system are running a small data centre in Melbourne, Australia, Cortical Labs says.
While the idea of growing neurons in the lab is not new, what Cortical Labs says it has done differently is to standardise a system that can be used more easily when connecting cell cultures to electronic interfaces, rather than requiring complex, custom-built lab setups.
Efficiency found in human biology
What previously required months or years of specialised lab work can now be done in hours or days thanks to its integrated platform, the company says.
Interacting with biological neurons in this way could make computing more energy-efficient and adaptable than conventional systems.
“Biology is incredibly energy efficient. We [humans] don't require huge amounts of data. Kagan said.
“I have a small daughter, and for her to learn what a dog is, she just needs to see a couple of pictures of a dog. Machine learning needs to be tens of thousands, hundreds of thousands, depending on what the task is. We can also deal with uncertainty, with noisy information,” he added.
Using human-derived cells could also have research applications. Because the neurons are grown from donor samples, they may reflect genetic traits, allowing scientists to study how cells respond to different treatments in a lab setting.
That said, traditional silicon-based computers remain far more effective at precise, fast mathematical calculations, Kagan said. Advances in current AI systems may be reaching practical limits, as they require ever larger amounts of data and computing power.
Instead, future systems are likely to integrate biological and silicon-based approaches to achieve capabilities that neither could deliver alone, the co-founder said.
“The future of computing is when we can leverage all of the tools that we have available to get the best result”.
Some experts agree that biological systems offer advantages such as low energy use and adaptability, but question how far current approaches can go.
“If you are only using a flat network of human neurons, I do not believe it would have any major advantages compared to traditional silicon-based systems,” Alysson R. Muotri, the Director of Sanford Stem Cell Education and Integrated Space Stem Cell Orbital Research (ISSCOR) Center at University of California, San Diego, in the United States, told Euronews Next.
He said more complex, three-dimensional brain-like structures, known as organoids, could offer greater potential, although these remain experimental.
Ethical questions around biology in computing
The use of human cells in computing raises ethical questions, although researchers say the level of concern depends on the complexity of the system.
Muotri said he does not see major issues with simpler networks of human neurons, such as those used by companies like Cortical Labs.
However, he warned that more complex brain-like structures could present challenges.
“The anatomic organisation of the tissue… can likely generate some kind of experience in a dish,” he said. “This might create some sort of consciousness… and some people might be uncomfortable knowing this.”
Such concerns, he added, could require new rules and oversight as the technology develops.
Kagan said Cortical Labs’ approach could offer ethical advantages, including reducing the need for animal testing and allowing greater control over biological systems.
“We find that this is a much better approach,” he said.
