US scientists plan to build computer from a plate of human brain cells

We know that scientists have grown a plate of brain cells and taught them to play table tennis.

And now, they actually want to use this disk of brain cells to build a computer?

In a paper published in Science Frontiers on February 28, a team of scientists described their plan to combine 3D human brain cell clumps (also known as brain cells) into organoids) into biological hardware capable of performing advanced computing tasks.

Paper address: https://www.frontiersin.org/journals/science/articles/10.3389 /fsci.2023.1017235

To put it bluntly, it is to develop a "biological computer" driven by human brain cells.

"While silicon-based computers are certainly better at numbers, the brain is better at learning," said John Hartung, a professor of microbiology at Johns Hopkins University.

This time, carbon-based organisms finally stood up?

Yes, those brain cells that can play table tennis

These scientists from Johns Hopkins University suggest that one day, laboratory-grown "Micro brains" are wired together to act as powerful and efficient biological computers.

Although AI seems to have taken over our lives now, these scientists believe that something called "organoid intelligence" or OI, driven by living human brain cells, It may one day surpass any AI and be more efficient.

After the birth of ChatGPT, there were warnings everywhere that humans would be replaced by AI. So now, relying on our own brain cells, are we finally going to get back in the game?

So, what exactly is this legendary "organoid"?

In fact, it is a 3D mass of biological tissue.

Magnified image of a brain organoid produced in Thomas Hartung's lab, stained to show neurons in magenta, nuclei in blue, and other supporting cells in red and green

Scientists named this field "organoid intelligence" (OI).

For nearly two decades, scientists have been conducting experiments on kidneys, lungs and other organs using tiny organoids - lab-grown tissue that resembles fully grown organs. No human or animal testing is required.

Recently, Hartung and colleagues have been studying brain organoids, pen-tip-sized spheres with neurons and other features that are expected to sustain functions such as learning and memory.

Team leader Thomas Hartung, a professor of environmental health sciences at Johns Hopkins University, said, "The technology to achieve biocomputing is now mature. We hope that some functions of the human brain , can be implemented as OI, such as making quick decisions based on incomplete and contradictory information (intuitive thinking)."

##Key information about "Organoid Intelligence" (OI)

This team also includes members from Cortical Labs, who are the ones who cultivated a plate A scientist with brain cells who can play table tennis. Sure enough, this is a familiar recipe and a familiar taste.

Using organoids for experiments has many benefits for scientists because they can bypass human or animal testing.

Hartung said in a statement: "Organoids open up research into how the human brain works. Because you can start to manipulate the system and do things that ethically cannot be done with the human brain."

Well, it feels like a dark scientist...

The human brain is stronger than a computer

If Let the human brain and the computer compete, who can win?

In Hartung’s view, modern computers are still weak in front of the human brain.

"Frontier is Kentucky's newest supercomputer, costing $600 million and occupying 6,800 square feet. Just last June, it surpassed the computing power of a single human brain for the first time. — but it uses a million times more energy."

Hartung acknowledged that although computers can process calculations involving numbers and data faster than humans, the brain is Much smarter when making complex logical decisions, like telling a cat from a dog.

Another example, when demonstrating the idea that "although silicon-based computers are better at numbers, the brain is better at learning", he gave the example that although AlphaGo was better in 2017 It defeated the world's top Go players in 2011, but it was trained using data from 60,000 games.

And a person would have to play five hours a day for 175 years to finish these games.

Silicon computers have reached their limits

Since 2012, Hartung has been reprogramming cells in human skin samples into embryonic stem cells. The brain cells are then grown and assembled into functional organoids.

Each organoid contains about 50,000 cells, roughly the size of a fruit fly nervous system. He now envisions using this brain organoid to build a futuristic computer.

Hartung said computers running on this "biological hardware" could ease the energy demands of supercomputing over the next decade, which are becoming increasingly The less sustainable.

"The brain has an incredible capacity to store information, estimated at 2,500 terabytes," Hartung said.

"We are reaching the physical limits of silicon computers because we can't fit more transistors into a tiny chip. But the brain is wired completely differently, with about 100 billion neurons Yuan, connected through more than 1,015 connection points."

16 years of research shows that the human brain's memory can store the entire Internet.

Because organoids share cells that enable our own brains to acquire and store information, brain clumps are particularly suitable for those who need to learn quickly and without much energy Computational tasks are consumed before the information is stored in compact neuronal connections in the brain.

As a result, they can also create complex brain networks that support more powerful calculations.

But what if the organoid intelligence, or biological computer, these scientists want to create fails?

Hartung said organoid intelligence does not pose a threat to AI or human brains grown the old-fashioned way.

However, it is time to increase the production of brain organoids and use AI training to overcome some of the shortcomings of our current silicon systems.

"It will take us decades to be on par with computers," Hartung said. “But if we don’t start planning for this now, it’s going to be a lot harder.”

Hartung said organoid intelligence could take decades to power systems as smart as mice. Provide motivation.

But by scaling up the production of brain organoids and training them with AI, it is foreseeable that this biological computer will have superior computing speed, processing power, data efficiency and storage capacity.

And, organoid intelligence could revolutionize drug testing research for neurodevelopmental disorders and neurodegeneration.

We wanted to compare normal brain organoids from donors with brains from autistic donors, said Lena Smirnova, assistant professor of environmental health and engineering at Johns Hopkins University. Organoids."

"The biocomputing tools we are developing, much like those that allow us to understand changes in neuronal networks in autism, can be done without the use of animals or contact with patients. Understand why patients have these cognitive mechanism disorders."

Ethical Torture

However, clumps of human brain cells are willing to be cultivated by humans A biological computer?

Will they be sentient, self-aware, or even feel conflicted?

Hartung’s team also thought of this. Some of its members, who have backgrounds in bioethics, are working to assess the ethical implications of working with OI.

And, if you want to build complex biological computers, there is an ethical minefield that researchers must cross - to build miniature human brain simulations, in brain organoids There is a limit to the number of cells with computing power.

But if you want to grow organoids suitable for computers, you need to expand from 50,000 neurons to 10 million.

And as computing power increases, these interconnected organoids will likely gain some form of intelligence, although they won't be fully sentient.

This brings us back to the age-old question: What is consciousness?

"As of today, we know that these organoids can mimic the neural oscillatory behavior of cortical development," said Alysson Muotri, a neuroscientist at the University of California, San Diego, and one of the paper's authors.

#These neural oscillation behaviors are brain waves.

In addition, under anesthesia, these oscillations gradually disappear, which is also similar to the human brain. "

"To bring the organoids within the realm of consciousness, we are doing some stimulation to collect PCI," Muotri said. PCI is a measure proposed by some experts to measure the level of consciousness of an organism or entity.

Lab-made mini brains can produce brain waves like premature babies, study finds

But will organoid intelligence emerge? When will it appear?

No one can know.

Hank Greely, a professor of law and genetics at Stanford University, specializes in the ethics of neural networks.

We may one day create a human neural organoid that can act like a brain and have some intelligence, he said. But even so, we still know very little about this question: What exactly creates consciousness in our brains?

"The number of neurons connected together doesn't make something intelligent. If I stacked a million cut rocks together, I wouldn't necessarily have Chartres Cathedral; I probably just have a bunch of cut stones," Greely said.

"Organoids are definitely not "mini-brains." They are not organized like a brain, they do not contain a large number of cell types like a brain, and they do not communicate with the body through inputs and outputs Continuous communication."

French mathematician and philosopher René Descartes believed that consciousness and body are two different things.

Faced with various moral crises, Hartung said, "There may not be a technology that will not have unintended consequences."

"Although it is difficult to rule out such risks, as long as humans control the input and output and the brain's feedback on their output results, they will not lose control. But if we give AI or OI autonomy, it will Something went wrong."

"In short, neither silicon machines nor cell machines can control humans."

In this regard, netizens said, "The real AGI is a new life form, whether it is on silicon or on meat."

"The autonomy of thought and the real desire to generate (not the pretended puppet like ChatGPT) are the evidence of life."

Paper Introduction

Let Let’s look at some key points from the paper.

Why develop OI technology

Scientists have always been enthusiastic about realizing biological computing.

Although the human brain is not as good as machines in terms of calculations, its ability to handle complex and open problems is far better than machines, and it consumes less energy and performs tasks more efficiently. high.

Organoid intelligence (OI) is the latest attempt by scientists to put this enthusiasm into action.

Johns Hopkins University researchers hope to create artificial intelligence using human brain cells. The team programmed human skin cells into a state similar to embryonic stem cells, allowing them to restore their ability to form brain cells and thus have basic functions such as memory and learning.

The time has come to develop OI technology

Regenerate human cells Programming stem cells and developing three-dimensional brain organoids from pluripotent stem cells makes OI possible.

In the past decade, brain cell culture has evolved from traditional monolayer cultures to more organ-like, organized three-dimensional cultures, Johns Hopkins University Center for Alternative Animal Testing and more The institution has produced this highly standardized and scalable brain organ.

Most of the cultured brain organs have active synapses, showing that they are comparable to the brains of premature babies cortical layers and shock wave patterns.

These organ cultures are also rich in a variety of cells involved in biological learning.

Cross-linking of OI with other techniques

Expose the cultured brain organoids to electrical and chemical stimuli, with input and output monitoring , verify whether learning occurs in it.

The large amount of feedback data generated by OI is analyzed by AI with powerful computing power, which requires correlating organ changes with specific output signals. In this process, OI It has built a bridge that integrates multidisciplinary tools such as machine learning, statistics, signal processing, and information theory.

Inferring the connectivity of organs is also an important task. Using tools from brain imaging and computational biology, it is possible to map the input and output of organoid nervous system connections and determine the connections between neuronal circuits. Whether an effective connection has been established.

It is necessary to provide fast, reliable, and scalable infrastructure for the storage and processing of OI data. It needs to be able to adapt to the large amount of data that each layer of biological networks can calculate and output, and it must also have the ability to dynamically change. .

The storage strategy adopted by the JHU team is mainly to develop a solution similar to the CERN Large Hadron Collider experiment, using complex triggers for real-time detection of events, and only those with potential discovery value Time will be preserved. In the future, by building an OI community, a powerful and reproducible big data infrastructure can be quickly established.

The further improvement of OI

First needs to improve the complexity of biological computing. In OI, organoids can play a role in the silicon interface and environment. The interaction function and the algorithm that optimizes silicon-silicon interaction can improve OI survivability.

We also hope to be able to input more complex biological signals to OI, such as input representations that are more similar to human vision.

Advances in molecular biology in synaptic plasticity are crucial to OI optimization, which can better express the effects of human brain learning by improving its growth conditions.

Vision for OI technology

Establish OI as a true form of biocomputing, ethically utilizing brain organoids.

Realize the revolution in biocomputing, overcome the limitations of silicon-based computing and AI computing, achieve faster decision-making, continuous learning, and improve the efficiency of performing these tasks.

Elucidate human wisdom more clearly and bring hope for the treatment of cognitive defects and other related diseases.

The above is the detailed content of US scientists plan to build computer from a plate of human brain cells. For more information, please follow other related articles on the PHP Chinese website!