2023-06-27 09:17:03
Artificial Brain Challenges AI! Can Brain Tissue + Brain-Computer Interface in a Petri Dish Beat Computers?
At the end of February 2023, Johns Hopkins University professor Thomas Hartung led a research team and published the research results of “Organoid intelligence” (OI), hoping to use brain organoids plus brain-computer interface to create a new biological computing technology.
Are we finally building artificial brains? OI and AI, who will dominate the future?
What is organoid intelligence OI? What is the goal?
Now in 2023, AI has demonstrated many amazing practical results; in contrast, OI is still just a fledgling project, and even the name is similar to that of the 2018 American “Natural-Physics” columnist, Physicist Buchanan’s article titled Organoids of intelligence is almost the same.
Organoid intelligence, Organoid intelligence, and OI are very new cross-field terms, combining the technologies of “brain organoids” and “brain-computer interface” at the same time.
Simply put, brain organoids refer to brain tissue produced by culturing or induced pluripotent stem cells (iPSCs) in a rotating bioreactor that mimics the in vivo environment. This technology, which sounds like it would only appear in a sci-fi movie, does already exist.
The earliest brain organoid was in 2007. The research team of Yoshiki Sasai and Takeichi Watanabe of RIKEN Brain Institute in Japan successfully cultured forebrain tissue from human embryonic stem cells. The first 3D brain organoid with different brain regions was published in the journal “Nature” in 2013, by Jürgen from the Austrian Institute of Molecular Technology. Koblich and Madeleine. Lancaster research team successfully established.
The emergence of brain organoids is of great significance in biological and medical research. This means that if scientists need to conduct brain-related research in the future, they no longer need to sacrifice experimental animals or teachers of anatomy to obtain human brains. Just make the brain we want.
Although the tissue on the petri dish is indeed brain tissue, in terms of size, function, and anatomical structure, the results so far are still far from our naturally developed brain. Therefore, to achieve the “intelligence level” required by OI, we must expand the existing brain organoids and make him a more complex and durable 3D structure.
To achieve the “level of intelligence” required for OI, existing brain organoids must be expanded into a more complex 3D structure. picture/GIPHY
And this brain must also contain cells and genes related to learning, and connect these cells with AI and machine learning systems. Through new models, algorithms, and brain-computer interface technologies, we will eventually be able to understand how brain organoids learn, compute, process, and store.
Is OI a type of AI?
Can OI be regarded as a kind of AI? It can be said that it is not.
The A of AI refers to Artificial. In principle, as long as it is artificial intelligence, it can be called AI. OI is the intelligence produced by artificially cultivated biological nerve cells, so it can be said that OI is a kind of AI.
But one faction doesn’t think so. Since the current development of AI is through digital computers, it is generally regarded as the wisdom generated by digital computers – AI and OI are like digital versus biological, and computer versus human brain.
OI has a chance to replace AI? What are its advantages?
As for why the accuracy and speed of computer calculations are much higher than that of the human brain, the main reason is that the design of computers has a purpose, which is to do fast and accurate linear calculations. In contrast, the neural circuits of the brain are a network of living connections.
The genetic composition of human beings and the environmental stimuli they receive every day are constantly changing the brain. Every minute and every second, our neural circuits are different from the previous state, so even if the single calculation speed is not as fast as that of computers, but The human brain has higher learning efficiency, scalability and energy efficiency. When learning the same new task, a computer even needs to consume 10 billion times more energy than a human being.
The neural network receives different stimuli. picture/GIPHY
From this point of view, at least OI has a higher advantage in hardware efficiency and energy consumption. If the advantages of AI and OI can be combined, AI software can be loaded on OI hardware to create a perfect computing system. It seems that it is not a dream.
But where has the development of OI come, and how far are we from this goal?
Possible barriers to OI and current developments
At the end of last year, Brett of the Australian brain science company Cortical Labs. Brett Kagan led the research team to make a petri dish brain that can play the ancient video game “Pong” – DishBrain. This DishBrain, which is composed of 800,000 cells and has a similar number of neurons to the bumblebee brain, takes more than 90 minutes to learn compared to traditional AI. It can master the game in just 5 minutes, and consumes less energy.
At this stage, institutions such as the Johns Hopkins Animal Replacement Center can only produce brain organoids with a diameter of regarding 500 microns, which is regarding the size of a grain of salt. Of course, such a size contains regarding 100,000 cells, which is already amazing. Although other research teams have been able to produce brain organoids with a diameter of 3 to 5 mm through a culture time of more than one year, there is still a long way to go before a brain organoid with a target cell number of 10 million.
To achieve the goals of OI, growing larger 3D brain organoids is a top priority.
OI improvement and multi-party integration
After all, brain organoids are still biological tissues, but unlike biological brains, they have a vascular system that can perfuse oxygen, nutrients, and growth factors and remove metabolic waste. Therefore, a more complete microfluidic perfusion system is needed to support brain organoids. Scalability and long-term stability of organ samples.
After the brain organoids are cultivated and confirmed to be able to survive for a long time, the most important thing is to analyze the data of the brain organoids’ information input and response output, so that we can know how the brain organoids perform biological computing.
Inspired by electroencephalogram (EEG) records, the research team will develop a 3D microelectrode array (MEA) for brain organoids, which can make the entire brain organoid elastic and soft in a way similar to wearing an EEG electrode cap The outer shell is covered, and large-scale surface stimulation and recording are performed with high resolution and high signal-to-noise ratio.
If we want to further and more thoroughly analyze the signals of brain organoids, superficial records are far from enough. Therefore, it is very important to obtain higher-resolution electrophysiological signals with minimally invasive invasive recordings. The research team will use the silicon probe Neuropixels specially designed for living experimental animals to further improve it into a device that is dedicated to brain organoids and can be used flexibly.
As the saying goes, learn from each other’s strengths and make up for each other’s weaknesses. To achieve OI, the use and contribution of AI is indispensable.
In the next step, the team will carry out brain-computer interface, and the implanted brain here is no longer a human brain, but a brain organoid. Through AI and machine learning to find out how brain organoids form learning memory and generate wisdom. In the process, because the data will be very large, the analysis of big data is also inevitable.
With the rapid development of AI, OI’s network voice has increased a lot, and it may have the opportunity to get more attention and research grants to accelerate research progress. What’s more interesting is that not only a group of people want to make AI more like the human brain, but another group of people want to make OI more like a computer.
The boundaries between biology, machinery and AI seem to be becoming more and more blurred.
OI = create “life”?
Speaking of this, it is unavoidable to make people worry, if one day OI really produces wisdom, will we create some kind of “life” out of thin air? This will inevitably lead to complex moral and ethical issues.
Although the research team also emphasized that the goal of OI is not to recreate human consciousness, but to study functions related to learning, cognition, and computing, but “what is consciousness?” This philosophical speculation has not yet been concluded.
Can an organism that knows how to “learn” and “calculate” be considered conscious? If a visual brain-computer interface is installed on the OI, will it find itself a biological computer trapped in a petri dish and slaughtered by scientists?
However, these problems are not only the problems that OI should worry regarding. With the development of artificial intelligence, GPT, Bing and other metal intelligences made of silicon will eventually face the corresponding philosophy and intelligence as they pass one following another intelligence and ability test. ethical issues.
Finally, Neuralink CEO Musk said (yes, it’s him once more XD), if human beings are not left behind by AI, they may have to rely on biochips and biotechnology to strengthen themselves. Faced with various choices of artificial intelligence, mechanical transformation, and biochips, what kind of life do you want to live in the future?
welcome to subscribe Pansci Youtube Channel Get more in-depth scientific knowledge!
1688003941
#difference #monkeysThe #Complexity #Cognitive #FunctionsA #Theory #Thousand #Brain #Intelligence #PanSci #总科学
