Would you connect your brain to AI?

Would you connect your brain to AI?

This is how Elon musk Brain-Computer Interface works

We are now connecting our brains to the cloud and in the coming years, we will likely be connecting our brains to AI.

Elon’s Musk company neural link released the groundbreaking progress that they have achieved with the “Brain-Computer Interface” (BCI) Technology.

Elon and his team expect BCI installation to be as painless and simple as an outpatient LASIK surgery without general anesthesia and with no stitches required.

We have all seen in the movies or wondered if there would be the day in which we could control matter with our thoughts. While this might have sounded like something from a sci-fi movie it is now a reality. And Neralink’s recent announcements contributes a major milestone towards the Brain-Computer Interface efforts that are making this possible.

You as I have probably experienced having a pocket of information or idea but it takes a long time to actually type it or write it into the paper. High Bandwith BCI will allow us to output information from our brain at a more efficient rate. An example of this is that of watching a movie and writing a story. Vision and hearing allow us to have a high bandwidth input into our brains as we can take a lot of information at a glance. We can absorb the sound, the visual effects, the emotion, and even the smell if we are in a more immersive experience. But if we were going to explain or tell the story of this experience, we would have to write it or speak it which is limited that is the main bandwidth problem.

In this blog I will be exploring:

  • The Progress of Neuralink’s pioneering work
  • The Process of how Brain-Computer Interaction works
  • The Plan of impact on humanity

The Progress

Neuralink has both a captivating business plan and pioneering work. The company’s brain implant is initially intended for the treatment of patients with cervical fractures and neurological disorders, allowing them to restore function. In the long term they will be available to the general population for enhanced capability or to enable AI enhancement of our brain.

In the public announcement, Elon outlined three main goals of Neuralink’s device:

  1. Increase by orders of magnitude the number of neurons you can read from and write to in safe, long-lasting ways;
  2. At each stage, produce devices that serve critical unmet medical needs of patients
  3. Make the procedure as simple and automated as LASIK.

Our brain is a fascinating organ, is a 3-pound soft beautiful organ that you can hold in a hand, is composed of 100 billion neurons and 1000 trillion synapses, encompassing everything we see, feel, hear, taste, and remember. Everything that makes us who we are.

In the near term, Neuralink aims to restore the function to patients who have suffered brain and spinal injuries, helping them to regain their ability to feel and regain some or all of their functions. Neuralink and Elon are not the only company who are trying to do this. There are many companies that are already bringing these types of Computer Brain Interfaces to market. An example is the deep brain stimulator that is used in patients with Parkinson's disease.

In the long term, The goal of Neuralink is to achieve a full “symbiosis with AI” according to Elon.

The Goal of Neuralink will be to overcome these barriers and increase speed, creating instantaneous, seamless access to on-demand knowledge, sensory experience, processing power, and meaningful output.

Let’s take a deep dive into how it works

The procedure:

Because the threads are so small, Neuralink had to develop a robot to perform the thread placement with high precision and accuracy. The incision made in the brain is just 2mm in size, which can be closed with crazy glue. The robots’ 24-micron needle is designed to precisely place threads and avoid damaging blood vessels.

No alt text provided for this image

They will be placed depending on their specific needs. For instance, in a patient with quadriplegia one array will be placed in the somatosensory region of the brain and three in the motor cortex.


According to Dr. Matthew MacDowugall, “We developed a robotic inserter that can rapidly and precisely insert hundreds of individual threads representing thousands of distinct electrodes, into the cortex in under an hour.”

The Team:

The multidisciplinary team includes Neurosurgery, Robotics, Materials, Electrochemistry, Micro-fabrication, Histology, Mixed-signal chip design, Optics, Metallurgy, Biochemistry, Machining, Firmware, Software, Neuroscience, Applied math, Software Security, FDA regulatory, Veterinary, Infrastructure, Clinical research. Only a team composed of these seemingly disparate fields can come together to create exponential advances in a focused target.

The Components:

One of Neuralink’s breakthrough is that they created the smallest electrode threads that exist. The size of these threads is one-tenth the width of a hair which is about 6 um in width, or the approximate width of a neuron. These are then inserted in the uppermost levels of the human cortex and interface with neurons.

No alt text provided for this image

1,024 of these threads attach into a chip that they call Neura link chip N1. this chip is embedded just below the scalp. Each chip collects and transmits 200Mbps of neural data, and up to 10 such chips implanted into a patient can allow for a grand total of 2Gbps wireless connection.

The wireless connection is then made via a blue tooth to a device that is mounted in the ear and connects the brain data to the cloud. The first interface is going to be an iPhone app. You will then be able to think and then your app will respond by moving a cursor or clicking a certain option.

No alt text provided for this image


The velocity of progress has been exponential, in the past two years, the size to performance ratio of the neural link electrodes has improved seven-fold. In these threads, they were able to fit the wires, the insulation for each one of the wires and the electrodes.

They increased the density and decreased the footprint of the threads and they have been able to increase the number of electrodes in each thread without increasing the width at the base of each thread.

No alt text provided for this image

The Team at Neuralink also designed a new method of packing these threads into the chip by using a lithographic process which is a process of printing that pretty much allows you to create whatever you can draw. So If they can draw the film in which the threads are placed, then they can fabricate it.

They also came up with a new way of making this thread film. The problem in the past with making these films was that you could not put so many of them in such a small film because you have to merge the materials that the threads are on and glue them together. Neuralink designed a new method to do this by actually making it together into one component. Creating a very small film that has over 1000 connections.

One of the impressive developments is that they were able to develop a system that surpasses the old way of developing these chips which used to take a year per chip from design to manufactured. The Neuralink team condensed this process into a 3-month process allowing Neuralink to have a completely different chip design every three months.

No alt text provided for this image

The Basic idea of these chips is the same. It needs to be able to sense a stimulation, then it needs to make sure that the electrodes are working fine or diagnose any problems with the electrodes, then it needs to amplify the signal, then convert that signal into usable information, then recognize patterns from the digital signal. And finally, it needs a power source.

Process

How does it actually work?

No alt text provided for this image

The threads are introduced into the brain near a neuron and are able to sense the spike of the nerve activity in an analog matter, or in other words, the thread or sensor wire will detect an electrical impulse coming down through the neuron, then the amplifier will take this analog spike in through a filter which will then convert it to a digital signal.

The 8 core computer will then transfer these spikes into digital signals. And this information is not only from one neuron and one electrode but from many electrodes in one neuron. And then from many neurons.

No alt text provided for this image

Just as in the very well known Electro Encephalo Graphy method; electrical activities are measured from the neurons. Now a bunch of electrodes from a single thread can read many impulses in real-time. Each line shows you the voltage signal in real-time as is coming off one of the threads. And if you then focus on one of those traces. The first thing you are going to notice is that there are these voltage deflections that happen periodically and these are the SPIKES. The chip can compute these collective spikes in only 900 nanoseconds which so far is thought to be faster than what it takes the brain to realize it happened.

No alt text provided for this image

Then this information is extrapolated into patterns and then organized into information that tells you about what is going on in the brain and what the intention of the thought is.

Neuralink also developed algorithms that can detect these spikes in real-time as they are happening. Then these algorithms analyze the data and create a spike roster. Each row represents one channel of recording and time goes from left to right.

Take for instance measuring the movement of the arm. If a subject moves their arm, the primary motor cortex would record these action potentials. They then built decoding algorithms that turn these signals into an organized pattern that can then tell you relevant information about the movement itself. And even if someone is paralyzed, even if imagined movement the cells in the primary motor cortex respond in a similar way as if you were making the move.

No alt text provided for this image


This decoding has been done in many labs before, but what they are trying to do is to increase the bandwidth so that it is not only one specific signal from a movement or input being recorded but a combination of inputs. For example, it would be able to not only read the signal for the intention of moving a mouse but also from the imagined intention to type on a keyboard. Potentially even the speech and even combinations of movement.

The Plans

So now that we have talked about how brain-computer interfaces work; we can take a dive into what Neuralink and other companies like it plan to do with these technologies.

And even though other companies and universities have made huge advances in the brain-computer interface, Neuralink took things to the next level.

In theory, you could decode the signal not only for the arm and legs but also for speech, and so it would have the capacity to decode information about doing more complex activities such as playing a sport or practicing martial arts.

And now, that we know that we can measure the electrical signals that occur when there is imagined movement which is similar to actual movement the question now is can we do the reverse?

One can help to wonder; Is this the birth of the jujitsu chip?

Meaning a chip that you could be placed into our brain that will allow you to learn jujitsu much faster. Or in other words, are we getting closer to having the ability to place information into the brain.? Yes, we can already do this through the sharing of knowledge via digital media and learning experiences. But that takes a long time. To get a black belt in jujitsu can take a lifetime. And these technologies are expanding on that capacity of the brain to learn and to grow, neuroplasticity and neurogenesis respectively.

In a way, this is the beginning of being able to send information to the brain. Meaning we know we can sense and organize the information of moving from the electrical activity of the neurons when thoughts occur but also the reverse is possible. And even though this might sound somewhat fantastical, the building blocks of this technology are already here.

No alt text provided for this image


And this technology is what is being used outside the brain for other devices such as cochlear implants, and in the retina to restore vision by other companies and researchers for many years.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6350159/#!po=10.0000

No alt text provided for this image

Neuralink is trying to expand on this; to read many more neurons at the same time and to send more information to neurons at the same time.

So it turns out that if you fire an electrode it causes the nearby neurons to activate and release an action potential, is like talking back to the brain and the somatosensory cortex has a very detailed spatial map to read these signals.

No alt text provided for this image

So if you touch a little part of the hand then the very specific small area of the somatosensory cortex lights up. Or if you touch two parts of the hand the somatosensory cortex detects two different spots or three spots. And there is research that shows that if someone is controlling a robotic arm and if the subject gets tactile feedback of when they are picking up a specific object with the robotic arm their ability to control that arm is improved.

No alt text provided for this image

Just imagine if you were trying to pick up a pencil with an anesthetized hand (this is what the current devices are doing). It would not only decrease the experience but also decrease the functionality. Neuralink is trying to create a closed-loop system in which the system can read from the brain but also send information to the brain.

You can also do it with vision. If you stimulate the eye with a little dot, then the somatosensory cortex will light up and a couple of dots will appear. And if there are two spots, then the visual cortex lights up with two spots.

No alt text provided for this image


So because the brain has a collection of many of these maps, not only spatial maps but also, maps that tell you about the color, the sounds, the orientation, the movement, and many other modalities; Neuralink’s goal is to create a device with such a density of sensors that can tap into the collection of these maps so that you can create rich visual feedback for the blind.

With the sensory input then the computer can decode it and then send a stimulus back to the brain and this could be used for conditions such as Parkinson's disease, Dystonia, OCD, Epilepsy, Depression, Chronic Pain, and Tinnitus among many others.

Not only are they hoping to improve motor and sensory functions but they speculate that cognitive functions can be improved. Neuralink is working on developing the characteristics of recording memories. For instance, if you are in a city and you are walking down the street, there is a spatial record of the movements and an imprint of each modality at that specific time. Meaning at each location there is a combination of a sensory map, a motor map, and auditory map, and a visual map and a sensory map. These are then combined to create a memory imprint. Some of this information is being recorded in the hippocampus. So the Neuralink team is working to understand how the hippocampus is able to store these imprints of different modalities into one memory.

No alt text provided for this image

The potential

I think the potential for this technology is incalculable. Definitely this technology promises to be a game-changer for people for different neurological diseases and even in our everyday life. With this technology people would be able to shorten the amount of time it takes for an idea to manifest into work or into actual physical creation.

Take for instance if you imagine the plans of a house, Then you sit down and you write the idea into a piece of paper. At that moment that was the first step that your idea became a reality or the first version of reality which would be a drawing, then you can go a step further to make the plans which would take longer depending on your expertise.

We used to use a piece of paper, a pencil and a rule to be able to draw out the plans, then we started to use cardboard models, computers and now we can create a real object with 3 D printing from a 3d drawing. Furthermore, with lithographic methods and CBI one could go from idea to 3d model in a faster time and more detailed manner. Especially, if there is a software to facilitate and accelerate this process.

No alt text provided for this image


The potential of this technology is hard to predict. It is clear that current hardware and software are being developed by many companies to use this technology to enhance communication, environmental control, movement control, locomotion, and neurorehabilitation.

The vast impact of human beings being able to use their thoughts to create actual work or manufacture objects could create a reality that is beyond our wildest imagination. In a way to try to predict it is a cognitively distant concept, meaning that some of it we will be able to predict and some of it we will not be able to because some of the possibilities that could be performed with this technology do not even exist in our cognition right now and can only surface once the progress is made.

This technology has the potential to enhance progress and quality of life for people currently suffering from many neurological diseases. And also trickling into other industries and even into our everyday life.

要查看或添加评论,请登录

Andres Zuleta MD MMM的更多文章

社区洞察

其他会员也浏览了