Australian Military Funds Computer Chip Fused with Human Brain Cells
Ultimate goal is to integrate with AI
This article is part of Future Explored, our weekly guide to technologies that are changing the world. Get articles like this one delivered directly to your inbox every Thursday morning.
The Australian Army is funding a project to grow intelligent “mini-brains” in petri dishes. The goal is to use these “DishBrains” to design better AI, eventually combining the two to create an AI that integrates the processing capabilities of human brain cells.
By creating the right conditions, scientists can grow stem cells into “organoids.” Organoids are three-dimensional tissues that resemble the structure and function of various organs (even the brain).
Organoids are not as large as the actual brain, but they contain neurons that can send electrical signals to each other.
For example, German scientists recently discovered that infecting brain organoids with the HSV-1 virus could potentially treat herpes encephalitis, a fatal brain disease. In 2022, researchers at Monash University in Australia did something unprecedented with brain organoids.
To do so, the researchers cultured human brain cells on microelectrode arrays. The microelectrode arrays are the same ones used in brain implants for sending and receiving neural signals.
In this game, players move their paddles up and down to obstruct the path of a bouncing ball.
This was done by stimulating electrodes on specific parts of the array to indicate the ball’s path and the position of the paddle. Brain organoids can stimulate specific electrodes to signal the paddle to move up and down.
To train this system, the researchers had to get creative.
Unlike animals, which can be easily rewarded for learning, it is not possible to give treats to a mass of neurons. Instead, the researchers used what they called the “free energy principle” to “motivate” the cells.
This theory is based on the idea that the brain uses what it knows about its environment to make predictions that minimize disorder. As learning progresses, correct predictions are made, and as the environment changes, the predictions are modified.
For the pong-playing DishBrain, researchers stimulated the organoids in a predictable manner when the ball made contact with the paddle (same voltage and same time across all electrodes).
When the ball passed the paddle, the electrodes were randomly fired to provide unpredictable stimulation.
During the study, brain organoids that received this stimulation had a “significantly increased” mean time to miss the ball. Organoids that did not receive such stimulation showed no increase in time.
We showed that by interacting with living biological neurons, we can force them to change their activity and produce something resembling intelligence,” said Brett Kagan, lead author and CSO of the biotech startup Cortical Labs.
What Happens Next?
In July 2023, Monash University and Cortical Labs announced that they had secured approximately US$400,000 (A$600,000) for further research on dishbrain through Australia’s National Intelligence and Security Discovery Research Grant Program. The purpose of this study is to explore solutions to a problem that affects AI algorithms but not the human brain: catastrophic forgetting.
While humans can learn one thing and retain that knowledge to learn another, AI does not.
Overcoming this problem and creating a system that can continuously learn, a system that we can easily learn from, could lead to the development of an AI that gets smarter and more capable the longer it exists.
With this grant, the Monash University research team will attempt to identify the biological mechanisms behind DishBrains’ continuous learning and replicate them in AI.
Says PI Adil Raj, “We plan to use this grant to develop a better AI machine that replicates the learning capabilities of biological neural networks.”
Nevertheless, building better AI is only a short-term goal.
Ultimately, the team hopes to replace conventional silicon computer chips with intelligent organoids to create advanced bio-computers. AI would then be built on the same foundation as the human brain and would be more efficient and flexible. Human neurons are self-programmed and infinitely flexible, the result of four billion years of evolution. We first emulate what the digital model tries to emulate.
The Big Picture
In February 2023, Kagan co-authored a paper in Frontiers in Science that gave the name “Organoid Intelligence” to this new concept of merging mini-brains and computers and detailed how it could revolutionize technology.
But the field is still incredibly new, and many hurdles stand between researchers and their vision of a continuously learning AI. One major hurdle is the need to increase the number of cells in brain organoids.
Thomas Hartung, a toxicologist at Johns Hopkins University and co-author of the Frontier paper, told CNN, “[Existing brain organoids] are too small. To increase the intelligence of the organoids, we need to increase this number to 10 million.”
Figuring out how to give brain organoids a robust vascular network could overcome the size problem. According to Hartung, it may be decades before we have a system with a mouse-like brain, but if researchers can fulfill the promise of organoid intelligence, the world will be a different place.
The brain is still no match for modern computers,” he said. The newest supercomputer, Frontier, a $600 million, 6,800-square-foot facility in Kentucky, is the first of its kind in the world. Last June, for the first time, it surpassed the computing power of a single human brain.
Source: Australian Military Funds Computer Chip Fused with Human Brain Cells
Injecting Algae Into Suffocated Tadpoles Brings Their Brain Cells Back to Life
Australian Military Funds Computer Chip Fused with Human Brain Cells