The Future of Brain Machine Interfaces: Symbiotic Intelligence vs Human Intelligence

We will explore what is Intelligence Amplification via brain machine interfaces (BMI), why it matters, and why there may be a future divide between humans who remain unenhanced, and humans who choose to amplify their intelligence by creating a synergetic symbiosis with Artificial Intelligence (AI).

Humans who connect with BMIs will be gifted with enhanced cognitive performance, and increased productivity in the workplace and beyond.

What is Intelligence Amplification?

The concept of Intelligence Amplification was first introduced by William Ross Ashby’s in his groundbreaking book titled Introduction to Cybernetics. The term then evolved to become what we now recognize as Augmented Intelligence, a subsection of machine learning that is designed first and foremost to enhance and improve human intelligence with the assistance of AI. The concept is to improve both human decision making, and the rapid access to information that humans have in order to enhance the quality of those decisions. This is where the current meaning of Augmented Intelligence ends, it is an AI that uses machine learning and deep learning to assist humans with actionable data, but there is no real-time symbiotic relationship.

This is where BMIs enter the picture, they will enable the enhancement of human cognition far beyond today's version of Augmented Intelligence.

Unlike our current access to data that takes place with computers, smart phones, or other devices, a BMI is inherently designed so that the internet, and the AI enabling access to the internet, can be accessed without an external device. The BMI will be implanted inside the human brain, and inherently becomes an extension of the human mind.

In other words, instead of relying on memory, or having to open a book, or visit a website, an enhanced human could have access to all of the information that is stored on the internet, and an advanced AI could feed the relevant data points to the human brain, enabling the human to be fully in control. If you've ever had a moment when you cannot remember a certain memory, or recollect a specific date, it is a frustrating experience. With Augmented Intelligence you could have perfect recall due to the AI system becoming an extension of your biological memory bank.

This type of intelligence amplification was further explored in “Man-Computer Symbiosis” a speculative paper published in 1960 by J.C.R. Licklider. This enlightening paper offers an early description of how humans must learn to control AI by forming a symbiotic relationship with the AI. As stated by J.C.R. Licklider, “To enable men and computers to cooperate in making decisions and controlling complex situations without inflexible dependence on predetermined programs'.

Machine learning is the secret sauce that ensures that a computer is of course not predetermined, nonetheless it doesn't tackle the issue yet of how we can access this symbiosis.

J.C.R. Licklider continued with this comment, “The hope is that, in not too many years, human brains and computing machines will be coupled together very tightly, and that the resulting partnership will think as no human brain has ever thought and process data in a way not approached by the information-handling machines we know today.” 

An early example of how this is being deployed can be seen in the world of chess.  While most people are familiar with Garry Kasparov's loss in 1997, to IBM computer Deep Blue, there is a newer and more interesting development.

While we've known for decades that an advanced AI system can easily defeat any chess player, what is more interesting is recent developments whereby an AI can be defeated by a human and AI team.  In this cooperative environment the team divides the tasks, the AI does the heavy lifting of massive computations, pattern recognition, and forward thinking. The human adds value by taking advantage of human intuition, and decades of studying the board.

While currently the human and AI team can defeat an AI, it remains unknown if this type of victory will remain constant moving forward. Nonetheless, this is a serious indicator that should humans be able to properly communicate, coordinate, and control an AI that is essentially an extension of their minds, that major problems that cannot be tackled by humans today, or by standalone AI programs, could be handled by a union of both.

One of J.C.R. Licklider's final comments clearly lays out the importance of designing BMIs capable of enabling real-time AI communication within the human brain.

“The other main aim is closely related. It is to bring computing machines effectively into processes of thinking that must go on in “real time,” time that moves too fast to permit using computers in conventional ways. Imagine trying, for example, to direct a battle with the aid of a computer on such a schedule as this. You formulate your problem today. Tomorrow you spend with a programmer. Next week the computer devotes 5 minutes to assembling your program and 47 seconds to calculating the answer to your problem. You get a sheet of paper 20 feet long, full of numbers that, instead of providing a final solution, only suggest a tactic that should be explored by simulation. Obviously, the battle would be over before the second step in its planning was begun. To think in interaction with a computer in the same way that you think with a colleague whose competence supplements your own will require much tighter coupling between man and machine than is suggested by the example and than is possible today.”

How does Intelligence Amplication Work?

Intelligence amplification via BMIs is still in its early days and is a work in progress. It must be understood that the human brain takes advantage of pattern recognition to understand symbolism and create connections between data. For example, if you see lines structured in a specific sequence such as the letter A, you can then recognize the symbol A. From there onwards you can have the letter form a pattern in your brain when you read the word APPLE. You can then recognize additional patterns when you read that AN APPLE FELL FROM A TREE. The human brain continues to make connections onwards from characters, to words, to sentences, to paragraphs, to chapters, and then to books and beyond.

The problem is the human brain does not have perfect recall, and this imperfect system causes pattern recognition systems to fail. Imagine what would happen if you could read an entire book and an AI system was able to form those pattern recognitions that are needed to instantly provide perfect recall. This would enhance the human's ability to work on an essay, to create products or services that rely on that information, or to simply have an intelligent conversation without any lapses in memory.

In other cases while in mid-conversation the human brain could instantly connect to the internet to in real-time locate information, and distribute or convey that information. Instead of having to watch a YouTube video multiple times in order to learn something, watching it once would be sufficient for perfect recall. The added advantage of additional pattern recognition systems is that the human brain could decode the video and audio faster than in real-time. This means the human could absorb the content of the video at speeds of 2x, 3x, or beyond.

Where Can I find Brain Machine Interfaces?

It is still very early days for this type of Intelligence Amplication. There are multiple efforts underway to develop various BMIs that could eventually evolve into this type of application. Most notable is Elon Musk's company Neuralink that is in the early stages of developing an ultra high bandwidth BMI to connect humans and computers.

Neurallink is working towards creating the first neural implant that will enable users to control a computer or mobile device anywhere they go. To achieve this Micron-scale threads are inserted into areas of the brain that control movement. Each thread contains many electrodes and connects them to an implant that is called the Link.

Even developers of a BMI system may not fully understand how it works on a micron neurochemical level. Due to the human brain's plasticity (ability to modify itself) it is actually the human brain that receives inputs and then learns by itself the necessary outputs for the BMI to work its magic.

Most BMIs uses a decoder to decipher the brain waves and patterns that are received by the human brain. This decoder uses various types of machine learning including deep learning to learn to decode the received information in an attempt to identify movement intentions, and wanted actions. By decoding these patterns it can best understand what the human brain is seeking to achieve.

It is a closed loop system where the user makes a motor intent by simply thinking, and the Neuralink decoder deciphers the intent. This translates thought into action which is then enacted in the world by a cursor or robotic arm. The human receives visual confirmation of a successful action and that neurochemical feedback trains the brain to more easily control the Neuralink. The challenge for any BMI company is building a decoder that is not too much of a learning burden on the end user.

Some of the issues with current BMIs involve latency, this is the time lag between input and output on both the human and the BMI side. Currently, Neuralink is working on fixing some of the issues that are involved with this issue as stated by Joseph O'Doherty, a neuroengineer at Neuralink and head of its brain signals team, in an interview.

“Step one is to find the sources of latency and eliminate all of them. We want to have low latency throughout the system. That includes detecting spikes; that includes processing them on the implant; that includes the radio that has to transmit them—there's all kinds of packetization details with Bluetooth that can add latency. And that includes the receiving side, where you do some processing in your model inference step, and that even includes drawing pixels on the screen for the cursor that you're controlling. Any small amount of lag that you have there adds delay and that affects closed-loop control.”

While Neuralink is the most popular example of a BMI, there are many other teams also working on fascinating projects. For example, researchers from the Howard Hughes Medical Institute have successfully enabled a BMI to type out the mental handwriting of users for the first time . The team deciphered brain activity associated with writing letters by hand to achieve the outcome. In this case with practice the brain learned how to strategically think about handwriting in a sequence that was then recognized by the BMI. The paralyzed participant was able to type 90 characters per minute, which is more than twice the amount previously recorded with a different type of BMI.

Another example includes a study with two clinical trial participants that have paralysis, and they used the BrainGate system with a wireless transmitter. Through the wireless transmitter, they could point, click and type on a standard tablet computer.

Amplified Symbiotic Intelligence vs Human Intelligence

We can imagine a world where some humans are augmented while other humans choose to be natural and fail to augment themselves. The danger behind this is that it will amplify the gap between wealthy humans with the financial means to augment themselves, and other humans who willingly or not remain unenhanced.

An employee that is enhanced will be able to achieve significant time savings by not having to second guess themselves, with an easy ability to instantly recall information or retrieve previously unknown data from the internet. An AI could quickly alert the human (or filter out) information that is irrelevant, fake, or substandard. The augmented human with perfect recall can pivot in how they accomplish tasks, and they could exponentially increase both efficiency and productivity.

Instead of typing text, or speaking outloud, the enhanced human could simply think and the text would magically appear on a screen. The time savings from this simpler version of a BMI would be significant. The BMI with the AI system may simply be implanted in the human brain, and wirelessly charged to external power sources, or be able to actually power itself from the same type of calories and resources that are built-in to the human body and brain. While it is super speculative, there may be nanobots that can cross the blood-brain barrier to generate a BMI.

An enhanced human may find that conversation with an unenhanced human is redundant, and boring. They may choose to associate themselves with other enhanced humans that wish to collaborate to launch businesses, write seminal papers, or become productive in other ways. An employer may choose to disregard educational background or experience, to instead focus on only hiring staff that has been enhanced.

Society could take different paths each leading to different outcomes. On one path there could be two types of humans who simply learn to co-exist.

Before BMIs reach this state the early developments are focusing on neurological problems that include the following:

• Memory loss
• Hearing loss
• Blindness
• Paralysis
• Depression
• Insomnia
• Extreme Pain
• Seizures
• Anxiety
• Addiction
• Strokes
• Brain Damage

It should not be forgotten that Neurallink's long term goal as stated by Elon Musk is, “To create a high bandwidth interface that allows humans to go along with the ride”. The implications are that if we successfully develop Artificial General Intelligence, this development inevitably leads us to Superintelligence. A BMI will be humanity's final solution for living in a world that features Superintelligence that is far more advanced than our current biological human brains. It remains to be seen how many humans choose to enhance themselves, in the meantime BMIs remain one of the most important developments featuring deep reinforcement learning systems.