Disclosure: I am not associated with NeuraLink in any shape or form.
As we’ve all have seen in the media and around the web, Elon has been having a rough few months. From sleeping on the factory floor in order to meet Tesla’s production objectives to consistently ticking off regulators and shareholders. It’s a wonder he has time for anything else in his life.
But then a month ago, he showed up on the Joe Rogan Podcast. While talking about things like Bonobos’ mating habits and taking half a hit of a well deserved legal blunt, he told Joe that his Neurotechnology company, Neuralink, is going to be announcing some ‘interesting’ news in a few months:
I think we’ll have something interesting to announce in a few months that’s at least an order of magnitude better than anything else. I think better than anyone thinks is possible.
— Elon Musk
What do we know so far about Neuralink?
For starters, the companies ambitions are nothing short of the most Sci-Fi Sci-Fi movie.
The company was founded in July 2016 by Musk and 8 others. Their short term goals are to treat serious brain diseases such as Alzeimers and Parkinsons and eventually get to enhancing the brain by “merging it with AI.”
The co-founders of the company are all highly reputable individuals that each bring a very special set of cross-disciplinary skills to the table to create Neuralink. Their skills cover everything from biology to neuroscience to computer science.
Here are the co-founders (other than Musk) involved in the company and a sample of their expertise:
- Ben Rapoport — Dr. Benjamin Rapoport graduated with an M.D./Ph.D. from Harvard Medical School. He also holds a Ph.D. in Electrical Engineering and Computer Science from MIT.
- Dongjin Seo — A neurotechnologist who has “created ultra-miniature, untethered, wireless neural implants (‘Neural Dust’) for brain-machine interfaces.”
- Max Hodak — According to his LinkedIn he “ studied biomedical engineering at Duke and worked in a lab studying multi-unit electrophysiology in rhesus monkeys to build brain-machine interfaces.”
- Paul Merolla — Has been the lead designer on more than 10 brain inspired chips including IBM’s TrueNorth chip.
- Philip Sabes — A researcher who through The Sabes Lab studies how the brain uses sensory feedback to maintain accurate and adaptive movement control. His lab also applies this research to the development of Brain Machine Interfaces (BMIs).
- Tim Gardner — A biologist who has worked on studying the neural activity in songbirds with the use of Brain Machine Interaces (BMIs).
- Tim Hanson — A Neurobiologist who taught himself about microfabrication methods and material science to develop some of the core technology at Neuralink.
- Vanessa Tolosa — A microfabrication expert, researcher of biocompatible materials and a Ph.d in chemical engineering.
Together with Musk at CEO, this looks like the dream team of the brain enhancement race.
What problem is NeuraLink trying to solve?
The best way to understand the problem is to first understand how we’ve been progressing into an increasingly intelligent species.
Elon says we‘re already cyborgs. The modern day human is able to answer almost any question you ask it within a matter of seconds. That futuristic device we all carry around connects us to an infinite pool of human knowledge. If someone makes a discovery, it can be shared with the human race within seconds on the internet. Combined with our devices holding access to this unlimited knowledge pool we‘re basically a giant inter-connected cybernetic organism composed of humans and machines.
Terminator inspired cyborg.
“your phone is already an extension of you, you’re already a cyborg.”
— Elon Musk
As our technology advances, our collective knowledge and intelligence only grows greater. A large part of this is due to us creating better communication systems to spread this knowledge across the network of humans.
We’ve come a long way from our ancestors in terms of developing communication. The humans’ progression of communication looks roughly something like this:
- Started learning to make sounds.
Sound waves.
2. Associated those sounds with the world around us.
3. Created language and passed down these sounds verbally as learning to be shared among others.
4. Drew symbols on surfaces to represent our world.
A caveman chiseling a rock.
5. Created written language.
Ancient Egyptian Hieroglyphics. Artfactory.com
6. Books. Available but scarce and accessible by few.
One of the oldest surviving books in the world. A Jewish prayer book. Dated around 840 AD
7. Mass produced books. Accessible to many more.
A large library of books.
8. Invented the telephone. Creation of another information sharing avenue.
Telecommunicatons network. waitbuywhy.com
9. Invented Radio and television. Methods of communication using more senses.
Early television.
10. Created Cell Phones — We’re no longer bound by physical location to communicate distantly.
Early consumer cell phone.
11. The Internet — Another form of communication, possibly the most powerful of them all. A giant network of knowledge with almost everything we know as a species. A collective brain.
Opte Project visualization of the internet as a global brain with pathways of neurons and synapses.
We’ve accomplished a lot as a species by overcoming obstacles to create these clever forms of communication. By creating more of these mediums, we’ve been increasing the rate at which we’re able to exchange our knowledge with each other. Creating machines and the internet was one of the biggest leaps in this regard. That is where our next hurdle lies: Taking our knowledge exchange abilities with machines to the next level.
Although we’re already cyborgs, the rate at which we can communicate with our machine counterparts isn’t very high. Musk says that we’re facing a bandwidth issue here.
“the communication rate between you and the cybernetic extension of your self, that is your phone, your computer is slow, it’s very slow.”
— Elon Musk
Now what if we could eliminate all the slow mechanical aspects of accessing the information available to us through machines?
For example, instead of typing something out with our hands, having the machine execute our request, then reading the information retrieved with our eyes, what if we could just think about it and access it instantly from the machine?
What if our minds could have access to a data storage device and instantly query it with our thoughts?
With these abilities we would be able to learn and remember exponentially better than using our existing physical methods of communicating with these interfaces. Not only that, it could help countless amounts of people who are suffering from brain related illnesses such as Alzheimers, where your memory is usually the first to go.
This bandwidth issue is one of the problems NeuraLink is hoping to tackle in the long run. One solution they’ve devised is to use a Brain Machine Interface (BMI) which will link from your cortex to a symbiotic AI to drastically increase your cognitive capabilities. Think Limitless but instead of a pill to juice up your brain, it’s an AI chip implanted somewhere in your body. An AI extension of yourself as Musk puts it. Ok, the Limitless example might be a stretch but I’m sure we’ll live that future some day.
When asked about the long-term goals of NeuraLink’s technology, Musk had this to say:
“It will enable anyone who wants to have superhuman cognition, anyone who wants. This is not a matter of earning power, earning power would be vastly greater after you do it. Anyone who wants can effectively do it, in theory.”
— Elon Musk
What is a Brain Machine Interface (BMI) and what exactly does it do?
Let’s first take a quick look at how our brain works and how scientists currently interact with it.
The average brain is composed of 100 billion neurons which form a information highway.
A neuron is made up of 4 main parts: The Soma (Cell Body), Dendrites, Axons and Axon Terminals.
In the diagram below of the single neuron, the Soma contains Dendrites which is where the neuron listens and receives signals from other neurons. There could be hundreds of these for each neuron. The received signal then triggers what’s known as Action Potential within the neuron which is like electrocuting it. As our neuron is electrocuted, it sends out more signals ahead to other neurons via it’s Axon Terminals which are making contact with other neurons’ dendrites thereby electrocuting them. The points at which the neurons connect are known as synapses.
A standard pyramidial neuron. Neurons come in many different types.
This system of communication operates similar to a computer’s binary language of 0s and 1s. The neuron is either excited with Action Potential or it’s not. The difference is that our brain constantly changes it’s neural connections allowing some neurons to be better connected to the rest of the network than others and vice versa. This change in connections is what we know as learning. When you learn something new, the neural network in your brain is creating action potential in neurons that never got much love before and in some cases generating brand new neurons.
Neurons experiencing Action Potential.
Now that we have an idea of how the neural network in our brain communicates, we can start understanding how to record and decipher it.
Interacting with the brain involves 2 categories:
1. Recording information we want from the brain.
2 . Inputting the information we want into the brain.
Currently, scientists and healthcare professionals are using tools like EEGs and fMRIs to analyze brain activity. These tools can provide great clues as to what’s going on in there but they are limited in the detail of data they can collect. The human skull is part of the problem in that it blocks much of the electrical signals we’re able to record. To really understand what it’s doing scientists need to get more detailed access to these signals.
Multielectrode Array which is implanted into beneath the skull.
One way to do this is to implant a multielectrode array underneath the skull. Sounds scary as hell, but it’s already been done. In the image above, the tiny needles on the electrode record the signals generated by the brain and are sent off to be deciphered by the machine interface.
Size comparison of a multielectrode array.
This is where we currently are with this technology and it’s quite amazing. It’s already shown the potential to change many lives. Watch the video below to see for yourself.
Knowing where we are with this technology now gives us a better idea of what we could expect from this upcoming NeuraLink announcement.
Now let’s add AI into the mix and talk about why Musk has been warning of “The Singularity” on almost any public forum he shows up on.
What is the singularity and why should we be worried?
T-800 Image by Dick Thomas Johnson
According to Wikipedia:
The technological singularity (also, simply, the singularity) is the hypothesis that the invention of artificial superintelligence (ASI) will abruptly trigger runaway technological growth, resulting in unfathomable changes to human civilization.
Above, we talked about how humans are essentially cyborgs because we’re part of this giant ever expanding cybernetic organism made up of humans and machines. Over time, we’ve been able to increase our communication bandwidth which has exponentially increased our abilities as a human collective.
We’ve come to the point where we are now creating a digital replica of our own cognitive abilities, creating machines that can “think” and do like us.
So how far are we humans in this process of creating the artificial us?
To compare, here are the 3 major parts of our brain that exist in us and what they do:
Three parts of the brain. (clinical-partners.co.uk)
Reptilian Brain — The oldest part, controls vital functions like breathing, heart rate, body temperature.
Limbic System — The mammalian part of the brain which is used for learning, memory, emotions and behavioral responses.
Neocortex — The “thinking” part of the brain which is involved in conscious thought, reasoning, language, motor commands and sensory perception.
If we were to try to measure how close we are, some might say that we’re in the neighborhood of the Limbic System. Artificial neural networks could be a good example of this. Their job is to take in data and learn from it as good as or even better than a human.
And that’s the part where it gets scary. These AI networks have instant access to what ever information we give it. When it comes to bandwidth, machines have exponentially higher than us limited only by their processing power. Imagine a machine making a discovery and sharing it pretty much instantly with the entire network of other machines. Every node in the network would be equally intelligent within seconds.
That makes you wonder… If we’ve come so far in such little time with our limited learning abilities, how much could a sophisticated AI network accomplish if given access to the wealth of human knowledge that already exists?
I find this question both full of great hope for our future and at the same time I’m thinking SkyNet.
This is why Elon Musk said:
“If you can’t beat it, join it”
-Elon Musk
When I started thinking about this, I realized, Musk has a point. He says once we’re able to successfully merge with our symbiotic AI counterparts,
“the outcome for humanity will be the sum of human will.”
— Elon Musk
I like the sound of that.