Cyborg computer combining AI and human brain cells actually works

Have you ever thought about a world where artificial intelligence meets human brain cells? This may seem far in the future, although technology is advancing and surprising as time passes. However, this scenario feels more real and present than ever.

Researchers basically created a brain using stem cells and connected it with a artificial intelligence, creating a kind of cyborg computer. Despite having some challenges ahead, it seems that we are about to enter a new era of technology, where humans and artificial intelligence mix in the way that many imagined decades ago.

The first hybrid computer

In the 2010s, researchers began developing brain organoids from stem cells. Therefore, we no longer know what these cells are capable of doing to revolutionize the area of ​​healthcare.

Advanced induced pluripotent stem cell (iPSC) technology is being used to create 3D models in an attempt to better understand complicated neurological diseases. This is super important, especially because mouse models, even when enhanced with humanized sequences, cannot fully replicate the characteristics of Alzheimer's disease. So, these brain organoids, which are basically like “mini-brains” created in the laboratory, can not only mimic the environment of a brain affected by the degenerative disease, but also include important parts, such as astrocytes, blood vessels and dysregulated microglia, which are crucial to understanding how the disease progresses. Furthermore, they are also very useful for researching new medicines.

Now, when we combine them with artificial intelligence, it creates a super interesting partnership in computational research. This union not only improves what we know about neurological conditions, but also indicates that we are entering a very exciting phase in machine learning technology. With this in mind, researchers from Indiana University Bloomington (UIB), in the United States, managed to achieve quite a feat by cultivating a cerebral organoid from stem cells and precisely combining it with artificial intelligence.

Initially with an accuracy rate of 51%, this would be a cyborg computer created and called Brainware, demonstrated constant progress in testing and achieved an accuracy of 78%, especially after rigorous training. Basically, these results indicate that the AI ​​computer may have the ability to learn and adapt when electrically stimulated. It is suggested that its ability is associated with neural plasticity, a property that highlights the ability of neurons to reorganize themselves in response to experience or injury, which is a desirable characteristic for computing devices.

Application in practice

But how did everything happen in practice? For testing purposes, the Brainware was subjected to a speech recognition task, challenged to identify sounds. With 240 audio snippets of eight speakers who pronounced Japanese vowels, he responded, and an artificial intelligence was trained to predict the speaker based on neural signals from the brain organoid.

In addition to his speech recognition prowess, he was also challenged to predict the Map of Hénon, a mathematical model known for being a bit confusing. Surprisingly, the performance of Brainware it was even better than the traditional linear regression technique, which is used to make predictions based on known information. To assess its general physical properties, other tests were carried out, including spatial information processing and transient memory. They were made by analyzing the response of ONNs (Artificial Neural Networks) to pulse stimuli with different times and voltages, which highlighted tremendous versatility.

Furthermore, the AI ​​computer was also part of a computing model composed of input layers, a reservoir, and an output layer. Scientists built this model by placing the cerebral organoid — which would be a “mini-brain” made from human stem cells — in a matrix full of electrodes. It functioned as the processing center, showing different brain cells and electrical activity. Signals were sent to the organoid by the input layer, which transformed temporal information into electrical patterns over time.

Results

In more complex challenges, the Brainware It surpassed previously known methods and demonstrated its vital dependence on the organoid. The lack of it resulted in a zero score in the regression analysis, which highlighted the critical importance of the organoid's contribution to the efficient functioning of the system. The remodeling of functional connectivity during training also clearly indicated the instigation of unsupervised learning. This means that, during training, the connections between different parts of the mini-brain (organoid) were altered, indicating that the training process was triggering a form of learning where the system adjusts on its own, without direct supervision for each individual. specific change. This ability to adapt is an interesting characteristic in the context of the study.

However, it is important to highlight that despite the promising advances, the human computer approach still faces significant challenges, as was somehow to be expected, after all, we are still in the discovery phase with artificial intelligence. The generation and maintenance of organoids, energy consumption by peripherals, the use of flat and rigid multielectrode arrays, and the lack of efficient data management tools represent obstacles to be overcome.

Still, given this scenario, we see a future where personalized and efficient systems, inspired by the brain, can be developed. With advanced brain-machine interfaces and improved data management software, the expectation is to achieve greater applicability and accuracy, thus shaping the next generation of artificial intelligence technologies.

AI with brain cells can be interpreted as the starting point for the development of more advanced and much more innovative biocomputing networks.

Watch the video on the Showmetech channel:

See also: Student develops Artificial Intelligence that translates sign language in real time

Sources: Freethink, labiotech

reviewed by Glaucon Vital in 29 / 1 / 24.