Computers Driven By Human Brain Cells!

The recent field of artificial intelligence can be described as very lively. A group of AI (Artificial Intelligence, artificial intelligence) products headed by Chat GPT (Chat Generative Pre-trained Transformer) have set off an upsurge on the Internet, with powerful performance  Constantly refreshing the cognition of netizens, many people think that the era when AI will change the world is coming.

The recent field of artificial intelligence can be described as very lively. A group of AI (Artificial Intelligence, artificial intelligence) products headed by Chat GPT (Chat Generative Pre-trained Transformer) have set off an upsurge on the Internet, with powerful performance  Constantly refreshing the cognition of netizens, many people think that the era when AI will change the world is coming.
Source Pixabay

However, some scientists pointed out that the current human-computer technology has reached a bottleneck, and the development of AI will be limited, so they proposed the concept of Organoid Intelligence (OI) and said that if OI can be realised in the future, it will be very promising.  It may make the "strong artificial intelligence" in science fiction movies a reality.


 AI, GPT and strong artificial intelligence

Before understanding OI, let's first talk about the AI ​​that is in the limelight. AI is artificial intelligence, which was first formally proposed at the Dartmouth Conference in 1956. It is a simulation of human consciousness and thinking process, and it is a means for human beings to understand the essence of intelligence. Professor Winston's definition of AI is "the study of how to make computers do intelligent work that only humans could do in the past".

AI, GPT and strong artificial intelligence
 Source 丨 pixabay

After decades of development, AI has been widely used in daily life.  For example, roles and affairs such as the intelligent voice assistant on the mobile phone, the face recognition that often appears in the authentication process, and the information push when surfing the Internet are inseparable from the participation of AI.


It also plays an important role in high-tech fields such as medical care, logistics, production, and transportation. Even in some areas where it is generally believed that AI is difficult to replace humans, such as painting and composition, AI has begun to be infiltrated.


painting titled "Space Opera House" won the Digital Art Award in an art competition held at the Colorado State Fair. This painting was created by a local designer named Jason Allen using artificial intelligence and then completed after polishing it with Photoshop

AI work "Space Opera House"

In 2022, a painting titled "Space Opera House" won the Digital Art Award in an art competition held at the Colorado State Fair. This painting was created by a local designer named Jason Allen using artificial intelligence and then completed after polishing it with Photoshop. The award of this work also caused considerable controversy in the local art circle.


Today's artificial intelligence needs a lot of data for learning and training. In the context of the big data era, the development of AI has become faster and more convenient. ChatGPT, which has recently detonated the Internet, cannot do without the support of huge data.


Invention of Chat-GPT and ChatGPT-4

ChatGPT is a text generation deep learning model trained on data available on the Internet, which has been updated to GPT-4. The knowledge of GPT-4 covers almost all subjects. It can generate grammatical and logical texts, and can also create paintings. It can even pass some difficult human tests such as SAT (Scholastic Assessment Test, commonly known as  The United States College Entrance Examination) and the Judicial Examination have achieved extremely high scores, which is amazing.


At present, all artificial intelligence is still in the stage of weak artificial intelligence. Strong artificial intelligence, also known as multiple intelligences, refers to artificial intelligence that has intelligence intellectual integrity and is capable of autonomous reasoning and thinking. It is considered to be able to produce perception and consciousness, and it is a new goal of AI development.  


There is a view that strong artificial intelligence may never be realised, but the emergence of the concept of OI has brought hope for multiple intelligences.  OI may be able completely a completely strong artificial intelligence by directly simulating the structure, thinking mode, and learning mode of the human brain.


 What is OI?

Many people believe that the emergence of ChatGPT marks the arrival of a singularity in the development of artificial intelligence, which will usher in a blowout development of the industry.  However, discerning scientists believe that today's computer technology and artificial intelligence technology are infinitely close to the bottleneck - because it is difficult for humans to install more transistors on chips.


In fact, for human beings, the human brain is the most complex intelligent system. The essence of artificial intelligence is also a simulation of the thinking process of the human brain. So can we directly create a "brain"?


In February 2023, Thomas Hartung's team at Johns Hopkins University proposed the concept of organoid intelligence, or OI, to use this theory to develop more powerful, efficient, and energy-saving forms of computers in the future.


Organoids refer to the cultivation of an organ model that is highly similar to in vivo organs or tissue sources through a 3D in vitro cell culture system. Although it is not a real organ, it has some functions of the original organ.


In February 2023, Thomas Hartung's team at Johns Hopkins University proposed the concept of organoid intelligence, or OI, to use this theory to develop more powerful, efficient, and energy-saving forms of computers in the future.
Source Pixabay


It is to use similar technology to cultivate the cell tissue of the brain in vitro, use it as the core hardware of the biological computer CPU and graphics card, and then connect it to external electronic devices through microelectrode arrays for training and calculation.


In December 2022, Cortical Labs, an Australian biotechnology start-up company, used human stem cells to induce the differentiation of human neurons and integrated these neurons with a multi-electrode array computer to form a fusion of a human brain and a computer. Then they also started to play a "ping pong" game through the microelectrode array, and the neurons learned the operation of this game in just five minutes.  An artificial neural network of similar size may take 90 minutes.



Since organoid intelligence is based on biological hardware, one of the biggest advances has been the reduction in power consumption.  The core hardware device of OI can almost be regarded as another brain, and the power consumption of the brain is extremely low.
 Source 丨 pixabay

Since organoid intelligence is based on biological hardware, one of the biggest advances has been the reduction in power consumption. The core hardware device of OI can almost be regarded as another brain, and the power consumption of the brain is extremely low.


The average power consumption of an adult's brain is only 20W, while the power consumption of most AI servers is usually around 10^6W.  Some top supercomputers, such as my country's Sunway TaihuLight supercomputer, consume 15.37MW (megawatts).  In addition, OI also has strong deep-learning capabilities.


OI learns better than AI

Because OI has the organisational structure of the human brain, it can realise "few-sample learning", and its ability to deal with complex problems far exceeds that of traditional AI. If you want to recognize a simple thing, the human brain only needs about 10 samples to learn it. In the early years, even if AI "learned" millions of samples, it still didn't learn.


AlphaGo, which defeated the world's top Go player Lee Sedol, was "fed" nearly 160,000 games of chess data. With such a training volume, a professional player would need about  37 years.


The human brain has a fairly complex neural network, just like a super multi-core processor. The thinking area of ​​the human brain is mainly concentrated in the cerebral cortex. There are about 14 billion neurons here, which can generate 1014~1015OPS (Operations per second, that is, the number of operations per second) computing power, which is the computing power of ordinary computers (1010 OPS), hundreds of thousands of times. With such powerful computing power, combined with a highly complex neural network structure, the human brain can evolve a deep learning model with a multi-layer structure. This set of models allows humans to efficiently learn and deal with various complex problems, which is millions of times more efficient than computers.


At the same time, after the human brain learns one thing, the experience, strategies, methods, etc. it acquires can also be applied to other things. For example, the thinking mode and methodology we cultivated by learning Go can also be used in other sports such as basketball and football. If AI wants to learn a new skill, it must start from scratch.


human brain learns one thing, the experience, strategies, methods, etc. it acquires can also be applied to other things.  For example, the thinking mode and methodology we cultivated by learning Go can also be used in other sports such as basketball and football. If AI wants to learn a new skill, it must start from scratch.
Source丨hippopx


There is no doubt that OI is a cross-age forward-looking technology, and there may still be decades to go before it can be truly realised.  Some scientists also raised their ethical issues, such as how to protect privacy rights, follow moral norms, respect human values, etc. The resolution of these issues also requires process and time.


In the future, if OI can be truly realised, it may set off a new round of frenzy in the field of artificial intelligence. By then, it may be possible to realise the strong artificial intelligence with perception and consciousness shown in science fiction movies.