You might say that, in the race to imitate human cognition, digital systems can go far, but maybe not far enough. In other words, what about the gap between digital neurons, manifested in binary or even quantum computing, and actual “gray matter,” the biological stuff that pulses and produces organic brain activity?
This idea has led to some interesting places, as scientists attempt to actually build brains, to create that organic matter with thinking capability, and as a visitor from the 1950s might say, that’s a horse of a different color.
Emerging Studies
I was reading about this sort of thing being done at Johns Hopkins University, where they have found these “three-dimensional cultures” to have some capacity for learning and memory.
“Right now, these organoids give us a human-specific, ethical way to study how learning and memory work. In the near future, they could help us test new drugs, understand brain disorders, and develop better therapies,” says Lena Smirnova, PhD, assistant professor in Environmental Health and Engineering, in an article posted to the school’s web site. “In the long run, this research also lays the foundation for ‘organoid intelligence’—biological computing systems that might one day complement traditional AI and even open new paths for brain-machine interfaces.”
More from Boston
I heard a lot more about this from Sri Sarma at TedX Boston, who covered some of the work being done on brain organoids in the defense industry, and, moreover, explained why it’s being done. Sarma is a Professor at the Institute for Computational Medicine in the Department of Biomedical Engineering at Johns Hopkins, and a real authority on this area of science.
She started off with some talk about drones.
“Autonomous drones are reshaping warfare,” she said. “They’re fast, precise, and often inexpensive. Built from commercially available parts, they can navigate, conduct surveillance, and strike in dangerous zones. As you’ve heard, they’re becoming increasingly autonomous. But they still struggle under uncertainty.”
Another problem, she suggested, has to do with the fragility of the systems on board.
“At the end of the day, these autonomous systems depend on fragile infrastructure,” Sarma said. “If the power goes down, the drone can fall out of the sky. If its GPS is spoofed, it can be tricked into navigating in the wrong direction. If communications are lost, it becomes isolated. And as environments grow more unpredictable, these systems reach their computational limits.”
In asking how to improve these aerial systems, Sarma noted that we already have a good model.
“These autonomous drones are highly capable, but they’re far from perfect,” she said. “The question is whether we can make them more adaptive under uncertainty—more resilient while remaining energy efficient. Nature solved this problem long ago with the human brain: nearly 86 billion neurons, no satellite link, no cloud infrastructure, just billions of simple units sensing, communicating, and adapting. From this network emerges something extraordinary.”
Mentioning that a brain is also ultra-efficient power-wise, running on 20 watts, she suggested that this is accomplished through the advantage of neuroplasticity, the agility of the brain to handle uncertainty, pivot under changing conditions, and compensate.
“Neurons strengthen and weaken their connections through lived experience, meaning they are not fixed systems,” she said. “They continually change with the environment—they learn, like a child acquiring language, and reorganize after disruption, like a stroke patient relearning how to walk. Machines don’t do this. They don’t continuously rewire.”
Then, too, she added, the chemistry of the human brain (and animal brains) is event-driven, tuned to stimuli.
“The brain spends energy processing information only when something meaningful changes in the environment,” Sarma said. “For example, when you put on your clothes in the morning, you initially feel them against your skin. A few minutes later, you hardly notice them at all.”
Machines, on the other hand, aggregate all kinds of real-time information and crunch it, whether it matters or not. Now, new transformer architectures are changing this. But Sarma’s point is well taken.
A Real Live Brain
Sarma stipulated, in describing modern science of this kind, that ultimately, copying something is different than working with it directly. Those professionals promoting digital twin technologies might beg to differ, but in any case, it explains why some teams are obsessed with having lab-ready biological organoids available.
Sarma explained:
“What if we could actually work with living neural systems—brain organoids? We interface with them. We observe them. And, most importantly, we guide them. That’s not science fiction. Labs today … are already doing this. DARPA has programs investing in exactly this direction. So the question is no longer whether we can—it’s how fast.”
The Recipe
Sarma described the process of “building” organic brains in a lab.
“We start with human stem cells and apply chemical signals that gently guide them toward becoming neural cells,” she said. “These cells find one another, organize, form networks and connections, and generate electrical activity. Over the course of weeks, they develop into three-dimensional brain organoids—living, responsive neural structures containing multiple brain regions and millions of neurons.”
She added that scientists can also use microelectrode arrays to observe the organoids, to get a kind of feedback loop going.
“These are microscopic systems that can sense and stimulate, and they’re flexible enough to sit directly against living tissue,” she explained. “What do they do? They record the organoid’s activity—they listen. They can also deliver controlled electrical pulses back into the organoid. In other words, they stimulate, record, observe, and guide.”
Making it Work
In talking about using this work in mission-critical systems, in defense, Sarma was careful to mention the dangers of error, using the term “control theory” to make her point.
“Control theory matters in defense not just for performance, but for accountability,” Sarma said. “If systems fail in commercial settings, you lose money. If they fail in contested environments, people can die. Control theory provides the mathematical tools to anticipate failure before it happens and to design robust systems that, if they must fail, degrade gracefully on our terms.”
A Combination
After enumerating various use cases for the technology, Sarma ended by suggesting that future technologies will merge the digital with the organic.
“The future of intelligence is not artificial intelligence alone,” she predicted. “It is something that emerges from biology and machines working together. It will not simply be engineered—it will be grown, guided, and used responsibly.”
I thought this was one of the more illuminating talks coming out of the Boston event. So much can be done with digital systems, but developing brain organoids is a very special part of our tech frontier, and not one, certainly, to be ignored. Stay tuned.
