The data storage capabilities of the human brain they are a complex topic and still the subject of scientific research and debate. There can be no precise measurement of storage capacity in terms of gigabytes or terabytes because the human brain works very differently than for example digital memories.
The human brain is an incredibly complex organ made up of billions of neurons interconnected. His skills rest on the training and strengthening of synaptic connections between neurons, through processes such as learning and experience.
How many terabytes of data can the human brain hold?
Many experts, however, have repeatedly made an effort to try to equate the human brain in some way with the behavior of the human being. storage digital. One of the first experts to have conducted estimates in this sense was the professor Robert Birge, award-winning chemist who has extensively dealt with the relationship between biology and electronics. Back in 1996, at the Syracuse UniversityBirge made a rough calculation and then refined it later, at the University of Connecticut.
Birge associated a neuron to a single one bit (in another article we talk about binary code, bits and bytes): a simple multiplication returns a value, in terms of storage capacity total human brain, equal to about 5 TB (Terabyte). In reality, the teacher has published a series of observations to estimate that a more correct value could be around 30-40 TB.
A subsequent study conducted by researchers from the Salk Instituteled by Terry Sejnowski, estimated that each connection in the brain could store 10 times more than hitherto known. So much so that one could even speak of Petabytes as a measure of how much data the human brain can store. The professor is convinced of this Thomas Hartung of Johns Hopkins University, which speaks of brain storage capacity of the order of 2.5 Petabytes, equivalent to 2,500 TB (we’ll talk about it later).
Other similar investigations place the brain capacity human between 50 and 200 TB. Although these are purely ratings theoretical, difficult to demonstrate with data and factual evidence, the values involved help us imagine the amount of data that each of us can potentially keep. Especially if the brain is trained to do it.
Artificial intelligence challenges the memorization skills of the human brain
If in the days of floppy disks we had just 1.44 MB of storage capacity available and these days Seagate presented the first 40 Terabyte capacity hard drives, the technological progress that has marked the storage segment has been truly swirling. With innovations continue on the technological side.
And if you think that, how can you do it even at home or in the office with the pattern RAID for example, the capacity of individual storage drives can be combined to create scalable storage solutions and with virtually unlimited capacity, the leap forward made in recent years seems even more incredible.
The advent of the modern artificial intelligences presents a real challenge to the memorization abilities of the human brain. Models trained on billions of parameters they are able to grasp the “essence” of the interconnections between terms and even concepts in an ever more precise manner.
Training, parameters used and memorization skills
A greater amount of training data it is particularly advantageous for artificial intelligence. The more data that is available, the more the model can learn from a wide range of examples and develop a deeper ‘understanding’ of pattern and relationships within the data.
The data quality training however plays an essential role: representative and coherent information ensures better learning of the model and contains any distortions.
While the human brain has an ability to limited storage, AI systems can use digital storage devices to store huge amounts of data. THE server some of the largest technology companies can store petabytes or even exabytes of data.
However, despite the data storage and processing capabilities of artificial intelligences, the human brain enjoys advantages significant in aspects such as creativity, processing complex information and adapting to new contexts. The human brain is capable of developing interdisciplinary connections, learning from emotional experiences, and applying knowledge in much broader ways than current artificial intelligences. Important news should come with the generative model GPT-5 of OpenAI which should be capable of generating and understanding multiple modes of information. One approach multimodal in artificial intelligence allows you to integrate and combine information from different sources. Think of the different sensory channels and data types (text, images, audio, video, positional information,…).
The next step: organoid intelligence
In this first part of 2023 we have also begun to speak with greater conviction about the concept of organoid intelligence. An article was recently published in Science describing the potential of bioinformaticawhich could turn out to be a sort of next-generation artificial intelligence in the future.
Artificial intelligence has long drawn extensive inspiration from the working mechanisms of the human brain. An approach that has proved to be very successful. But what if instead of trying to make AIs more like the brain, we went straight to the source?
Thomas Hartung explained it by revealing that teams of scientists engaged in multiple disciplines are working to create revolutionary biocomputers in which three-dimensional cultures of brain cellscalled “cerebral organoids,” serve as biological hardware.
What are brain organoids and how do they work?
The brain organoids they share key aspects of brain function and structure such as neurons and other brain cells essential for cognitive functions such as learning and memory. Being 3D structures, the cell density of the culture is increased by 1,000 times compared to a planar solution: it is possible to form many more neural connections.
“While silicon-based computers are certainly better at numbers, the brain is much more effective at learning“explained Hartung. “The brain not only learns better but is also more energy efficient“. The expert cites as an example AlphaGo, the Google DeepMind artificial intelligence which – trained on hundreds of thousands of games – was able to beat all humans, even the champions of the various games. Here she is amount of energy expended to train AlphaGo is more than it would take to sustain an active adult for at least a decade.
“Brains also have an extraordinary capacity to store information, estimated at 2,500 TB,” Hartung added. “We are reaching the physical limits of silicon computers because we can’t fit multiple transistors into a tiny chip. But the brain is wired completely differently. It has about 100 billion neurons connected through over 10 15 connection points. That’s a huge difference in power compared to our current technology.”
The structure of biocomputers with organoid intelligence
The goal will be to use brain organoids grown in the laboratory to make them become an actual organoid intelligence. THE test are continuing in recent weeks: Hartung explained that the current cerebral organoids are too small. Each contains approximately 50,000 cells; to arrive at an organoid intelligence, 10 million are needed.
In parallel, the authors are also developing technologies for communicate with organoids: in other words, to send them information and read the product. The study authors intend to adapt tools from various scientific disciplines, such as bioengineering and machine learning, as well as design new ones stimulation devices e registration.
“We have developed a brain-computer interface device that is a kind of EEG plug for organoids, which we presented in a paper published last August. It’s a flexible shell that is densely covered in tiny electrodes that can both pick up signals from the organoid and transmit signals to it,” Hartung said.
The authors predict that OI will eventually integrate a broad…