6: The obligatory Neurallink post
Or why I am so gung ho about the future of Brain machine interface
Everyone has said something about Neurallink at this point, so I figured I might as well riff about it too, share my thoughts, see what my stream of consciousness will make of this whole thing.
I
Filling a blank page with words or using speech to impart thoughts and concepts is such a slow, inefficient process of transferring information from one mind to another. I think about what to say then brain outputs it to fingers that type on a keyboard in order to form sentences that can be scanned by your eyes. Ughh, boring process. We’ve made significant process in facilitating the transfer of data over the millennias through global trade and the internet, leading up to us becoming cyborgs, unable to live without our newly made artifical limb, the phone. Even with all that progress in data transfer, the process is still slow and a lot is lost in translation. But things are changing, finally.
To understand Brain machine interfaces, we need to start with how the brain works. The brain is composed of 100 billion neurons in total, 20 billions of which occupy the part of the brain we call the cortex(the rational thinker). We’ll leave our limbic system(fears, emotions..) be for now. Think of the brain as a huge forest, and of trees as neurons. Now, these neurons are connected through a network that spans the whole body by means of a nervous system which extends to the spinal cord, down to the rest of the body. Me writing these sentences now means that electricity is whizzing in my brain. Why electricity? information in the brain flows by means of electrical signals. A neuron( a tree) is composed of a dendrite at one end and an axon at the other(roots), and the soma( the trunk) at the center. When you think of lifting your hand, a neuron fires or action potentials: Action potential-ing means a neurotransmitter has been fired through synapses at the ends of two communicating neurons, ready to transmit information in the form of an electrical signal from one neuron to the other, and the process is reiterated all across the neurons that are going to “lift the hand”. So electrical spikes are indications that information is traveling in the brain. Okay, if the information in the brain is mere electricity, shouldn’t we be able to capture these electrical signals, decode them, and then use them to control things? Yes we caan!
Brain machine interfaces(BMIs) sound like the stuff of scifi, but they go way back to 1957. Since 1957, Progress in BMIs was annoyingly slow. Sure, we (by we i mean them,the engineers/scientists) figured out how to restore hearing loss( chochlear implants), partly restore vision( though foggy, still better than pitch darkness), we made two distant brains send information to each other telepathically (a pretty rudimentary form of telepathy) to play a game, we made a qudrapelegic guy play Tetris with his mind, and helped people with parkinson’s disease tbrough deep brain stimulation, and the list goes on. As for what we currently know about the brain, one neuroscientist says: “ If the brain is one mile, we only walked for 3 inches” Not a lot. The techs we use today to understand the brain and its functions are measured along three metrics( resolution, Invasiveness, and temporal accuracy),and they range from CT scans, FMRI machines, to head covered with electrodes all over thingy( think the movie transcendence), and other relatively invasive techs. But understanding the brain is akin to putting our ears to the wall of a factory and trying to understand what every machine inside is doing. The techs we have now can only tell us that there is a hodgepdge of noise going on inside, but can not exactly determine what purpose each machine is serving. Fortunately, understanding the brain is a scientific problem, and we don’t need to understand the brain to be able to interface with it. Interfacing with the brain is a an engineering problem(s) and we can solve it, and the result of solving that problem, hopefully, will tell us more about the brain.
Here comes the role of Neuralink. Neuralink’s mission is to interface brain and machine at a high ultra bandwidth. The company started with 160million$ seed investment, 100 million dollars of which were Elon musk’s money, so it is pretty much an Elon’s company. Elon looks at end results of his companies as streams of probabilities,so any one comapny of his follows the following strategy: Find an industry where progress is slow or non existent, but if achieved, its end result will ensure a better future for humanity, light a match in that industry( by showing that it is possible to innovate in the field and bring to market an affordable, impressive product), let other big players in that get excited about that future, which leads to them joining in, and marshaling their resources(talent, funds) to that specific end,and the probability of that future happening increases, ensuring hence a better future for humanity. I know, cool strategy right.
II
Okay, getting into the weeds here. The problem with old BMIs was the low bandwidth. In order to record information from the brain and into the brain, you need to place electrodes so close to a neuron such that it could record the electrical signals( data) that surronds it when it is action potentialing. The number of electrodes that we could insert into the brain at any one time has been increasing slowly, and we’d rather it follows a Moree’s law curve, where the number of electrodes we could put on our chip doubles every couple months or so, sorta like what happened with the advent of integrated circuits. Neuralink bumped up the number of electrodes to 1024 electrodes, aka channels. Still a low number compared to the bandwidth we wish to achieve (reading from one million neurons). We need to collect information from as many neurons of the 20 billion neurons in our cortex as we can. So that is one engineering problem we need to figure out. Next comes the problem of how to place the implant in the brain without causing any trauma or brain injuries, and even if successful, we don’t want the patient with the implant to have side effects of any sort. In their first demo, Neurallink revealed a sophisticated robot that could implant the electrodes(4 micrometer in diameter) very delicately, avoiding any blood vessel that might come in its way. We also beared witness to a bluetooth chip whose tiny electrodes extends down into the brain,arriving at the area adjacent to the ear, and connected to a device placed behind the ear, outside the brain. Now that was cool but in the recent demo, Neurallink did away with the behind the ear chip, and replaced it with a chip the size of a coin that is impanted 4mm into the skull, replacing that part of the skull, and then the opened area of the skull gets sealed shut with glue, leaving only a tiny scar that is invisible under your hair. The electrodes emanating from this chip don’t go far into the brain, only a tiny distance, bit still tapping into a large potential of the cortex. Suppose we have this coin size implant into our brain and nobody is aware of that. Neat but what will it be good for? Neurallink’s short term goals with these kind of chips are to cure diseases that come with old age like amnesia, dementia, and paralysis. We saw in the recent demo how the chip that was implanted into a pig didn’t produce any side effects or affected the pig’s behavior and health. We also saw how it accurately predicted legs positions of the pig on a treadmill in real time, in addition to a live feed of the info the chip recorded whenever the pig’s snout touched things.
When asked about their hopes for this tech, The researchers/engineers at Neurallink had answers that ranged from curing blindness, decreasing suffering in the world, to saving memories, replaying specific memories, uploading memories to another bodies or robots, telepathy( conceptual and consensual). And eventually the weird future of merging with AI, or having that tertiary layer as Elon calls the chip serves as an AI extention for yourself. AI will eventually surpass human intelligence, there is no doubt in that. We saw how Gpt3 creeped the hell out of people weeks ago, and the progress is accelerating. Elon knows that very well, which is why he previously founded Open AI, the non profit open source company that ensures AI alignment with human goals, and he made it public. Open AI publishes every new breakthrough for the whole world to use. Why? Elon thinks that if the power of Super intelligent AI is concentrated in the hands of a few, humanity will be at a disadvantage, but if we have everyone on earth develop super AI, the power will be distributed all over the world and no one will prevail. “If only superman is from Krypton there is a problem, but if everyone is from krypton, we will be fine”. When Neurallink finally achieves the aimed ultra high bandwidth, the merging will be possible. If,on the other hand, the communication between you and the AI part of yourself is low bandwidth, the AI will rebel against you, and make decisions on its own, But if you become one through high bandwidth communication, You’ll make decisions together. Rejoice!
III
I wasn’t surprised when i learned that 6 members of the team of 8 that founded the company in 2017 left the company, and the reason behind it was that those people wanted to take things slow, and do science in the most literal sense of the word; they wanted to get the science right first and then apply what they learned. While that method of doing things is nice, in Elon’s world, things can’t wait for long, and there is a need to move fast and break things. Elon sees things from an engineering perspective, and he puts so much pressure on his employees to meet very strict deadlines. It worked out well for him in Tesla and Spacex. He is an innovator and if you cannot innovate ways to doing things faster than anybody else, you don’t have a place in Elon’s world. Good for Elon.
Let’s think a little bit about the future of Neurallink. So in the future, I’d be able to experience what the rich are experiencing on a daily basis, i mean, if all what they are experiencing comes from the senses and gets recorded as electrical signals in the brain, it would be easy to save those experiences and let anybody in the world experience them. Cool. what else? Empathy would be heightened. What about the world of sports: if you got injured in a game, you could still play the game using an artifical body of some sort, from the comfort of your own house, maybe; what is a skill but a collection of recordable electrical signals in the brain. What will happen to good ol’ conversations. People would have rich conversations without the need to utter a single word, through telepathy. You’d see two people exchanging conversations silently with occasional smiles and giggles. For telepathy to happen you’d have to consent or grant access to your thoughts to other parties, or else they won’t be able to. In this vein, a lot of security issues come to mind: what if someone can hack into my thoughts? or more scarily, what if they can download thoughts into my brain, and i would suddenly change my mind about things without knowing how it happened. Now these are serious issues and should deinitely be addressed going forward. Elon says that security is top priority at Neurallink, but i don’t know man, lot of bad guys out there. The thought of you being able to upload your self to the clouds is unfathomable. Makes you think about the whole meaning of You. So I died and my existence now lives in a robot, am i the robot now? Or maybe you can live in the head of someone you know( what!) What would that be like? Imagine like a musuem that contains the experiences or memories of everyone that lived, and you can buy a copy of any life of your choosing, and that life for couple hours. INSANE. The potentials are endless. Imagine being able to impart all what’s in your head from ideas, to animation, to films,to engineering projects, directly onto a computer screen without the need to learn the skill of how to do so. What about knowledge? Currently, we have all the knowledge of the world at our fingertips, thank to our internet connected mobile phone, but you still need to make an effort to find out info. What if you can learn things by merely thinking about them; say for instance that i wanna know how a rocket travels to the moon; i think about it and suddenly i know it, like immediately. Amazing. I am walking down the street and to my surprise, a gang of ten have plans of picking a fight with me. I go on a full rambo mode by mere thinking about what rambo woyld do in my stead. Enough with the future, my head is about to break. But yeah, I am definitely looking forward to merging with the machines. Transhumanism is my thing, it turned out.
Phew, feels cathartic. I can finally sleep now, knowing that i said everything( what I remembered) about Neuralink. Hurrah,I am free.