Computers that run on human brain cells … It sounds like something that only appears in science fiction, but in the eyes of scientists in the field of organ-like intelligence (OI),
Combining neurons with chips is not only expected to improve the computing power of existing computing systems, but also help to understand and treat brain diseases.
.
According to researchers at Johns Hopkins University,
A "biological computer" driven by human brain cells may be developed in our lifetime. They predict that this technology will multiply the capabilities of modern computers and create new research fields.
. In a recent article in Frontiers in Science, they expounded their "organ-like intelligence" plan and the possible challenges.
Professor Thomas Hartung, who led this work, said, "Computing and artificial intelligence have been pushing the technological revolution, but they are reaching the upper limit.
Biocomputing aims to reduce the resources needed for computing and improve its efficiency, so as to break through our current technical limits.
. "He thinks that the brain is still unmatched by modern computers. The latest supercomputer "Frontier" located in Kentucky, USA, is a device with a value of 600 million dollars and an area of 6,800 square feet. It was not until last June that its computing power surpassed that of a single human brain for the first time-but it used 1 million times as much energy as the human brain.
What is an organ-like?
In the past two decades, scientists have been trying to establish an in vitro culture tissue model (organ-like) that can simulate internal organs to carry out experiments based on kidneys, lungs and other organs without the help of human or animal experiments. The research team of Johns Hopkins University has been studying brain organs, which are 3D brain tissues with neurons and other brain characteristics, and are expected to maintain basic functions such as learning and memory. This "brain in a vat" provides unlimited possibilities for scientists’ research. As Professor Hartung said, "
This opens up the study of how the human brain works, because people can start to manipulate the system and do some experiments that cannot be done with the human brain morally.
. "
Dr Lena Smirnova, the first author of the article, pointed out: "
The combination of organ-like and artificial intelligence can make us take a step forward in finding drugs to treat Alzheimer’s disease or helping us treat children with developmental or learning disabilities without spending millions of dollars on preclinical experiments and unsuccessful clinical trials.
. Dr Smirnova’s goal is to turn these human-like organs into alternative research models for environmental toxicology research. According to her, animal models can’t accurately reflect how substances affect brain function and are expensive, so many chemicals in the environment have not been tested at all.
Image source: 123RF
The advantages of quasi-organs in economy and experimental performance have been favored by the pharmaceutical industry. Dr Hans Clevers, an executive of Roche, revealed in an interview with Nature that the organ-like model has been used in preclinical research in many laboratories, and the pharmaceutical industry is seriously facing up to the use of organ-like models and using them in every stage of drug development.
How to make brain organs perfectly reproduce brain function
Any kind of organ should accurately reflect the structure and function of its corresponding human organ, but brain organs have some unique limitations compared with other types of organs.
For brain tissue, its main functional feature is the ability to process information in the form of learning and memory. However, the cells in a Petri dish cannot display their learning ability in a meaningful way, which leads researchers to be unable to intuitively understand how substances (such as drugs) affect brain function.
. Therefore, researchers are trying to make these in vitro cultures reproduce the function of human brain. They believe that organ-like intelligence can make up for this defect, and it can also reveal the mechanism that supports learning and memory, and provide potential treatment strategies for diseases such as Alzheimer’s disease.
“
Imagine that if we can prove that an organ-like model in vitro has the learning ability, then we can compare the healthy organ-like model with the organ-like model produced by the brain cells of Alzheimer’s patients, and perhaps we can find something to restore the learning quality.
. "Professor Hartung explained. Professor Hartung’s analysis is not without evidence. The evidence he cited is DishBrain, which is a closed-loop system composed of 2D neuron cultures connected to a computer that can send and read cell signals at the same time. In October 2022, DishBrain’s creative team published a paper in Neuron magazine, detailing how they successfully taught neurons to play the computer game Pong in the 1970s. Dr Brett Kagan, a member of DishBrain’s creative team, is also one of the authors of the paper on organ-like intelligence published in Frontiers in Science.
At present, the 3D brain organs in Dr. Smirnova’s laboratory are still in the early stage. Compared with the real human brain, this kind of organ is very simple, with neither too many brain folds nor more complicated structure.
but
Since a 2D model like DishBrain has learning ability, a 3D organ model is likely to have more functions, because its structure is more complicated.
.
The researchers intend to reflect their learning ability by characterizing the gene expression patterns and electrical signal responses of human brain organs. If human brain organs really have all the elements of the learning ability system, it is time to connect them to a computer and put them in a learning environment-thus establishing organ-like intelligence. However, even so, this is not enough to achieve the goal of truly rebuilding brain function. Learning and memory is a complex process involving many brain regions and cell types, and I am afraid that it is impossible to accomplish such delicate and complicated functions only by neurons.
In order to transcend the formation of short-term memory, human brain organs may need to form a larger structure with different levels and regions. In addition, it also needs immune cells, which are very important for the connection between neurons, and some structure similar to blood vessels or perfusion system to help penetrate nutrients into organs.
. Dr Smirnova said that all these aspects have made progress, but it is still too early to produce an organ-like body that can completely reproduce brain function.
Image source: 123RF
Challenges in semantics and ethics
Technical constraints are not the only problem that must be dealt with in this field. The most fundamental level is the challenge of semantics: when the paper on DishBrain was published, many neuroscientists refuted his viewpoint of establishing perception, goal-oriented behavior or intelligence in neurons in Petri dishes. Some of them wrote articles in Neuron magazine to explain their concerns, pointing out that the use of these terms is not only inflammatory, but also misleading.
“
It is claimed that cell cultures cultured in closed-loop systems show perception and intelligence, which may affect the public’s understanding of the nature of perception and intelligence, and may lead to ethical debates intensified by misunderstanding.
. The dissenting researchers retorted in an article in Neuron magazine. In addition, they further point out that the application of intelligence and perception of neurons cultured in vitro in DishBrain’s paper is not based on any established consensus on the definition of these terms. At the same time, the results of this study have also been criticized. Some scientists think that they are too thin to prove the correctness of their conclusions.
DishBrain’s research team acknowledges that other scholars’ concerns about the language used in the article are reasonable, but they deny that they have exaggerated their research results in the article. As a follow-up progress, they added that the research team has contacted ethicists to understand the meaning of the term "perception" in the context of neurons cultured in vitro.
Image source: 123RF
The semantic problem of perception is only a small part of the problems that scholars in the field of organ-like artificial intelligence have to solve. Most ethical issues about artificial intelligence are concentrated in the field of consciousness, and the definition of consciousness is not clear.
If brain organs can produce even a little consciousness-like characteristics, will they also have the ability of pain and suffering? Do they also have rights?
In this regard, researchers say that a very big challenge they face is to do it in a very ethical way. In order to avoid moral and ethical dilemmas, they regard bioethicists as a key part of their OI plan. In addition, they also proposed to establish a set of common terms, formulate best practice guidelines, and study the neural basis of consciousness.
Most of this OI paper focuses on exploring more far-reaching theoretical applications in this field, such as using brain cells to run computers to make them more efficient, but scientists believe that this scenario is unlikely to be realized in the short term. As for other advances in the field of OI, such as testing the effects of drugs on the developing brain, it may not be so far away. As the researchers said in the article, there is still a lot of work to be done before the brain-like organs can completely reproduce the characteristics of the brain and even form long-term memories, but the exploration along this road will be transformed into the practical application of OI.
Professor Hartung revealed that his team’s next plan is to use such organs for drug development or toxicology research. According to their estimation, they will have a replicable system to simulate learning in one or two years. Then, by cooperating with scientists who study autism and Alzheimer’s disease, they plan to establish a system to study neurological development and degenerative diseases. At the same time, he and his team members welcome public opinions. Perhaps the development of this technology is very terrible for the public, so he hopes that people will have a comprehensive understanding of the benefits and challenges of OI from the beginning.
