It's a long read, but fascinating. The analogy of comparing the brain to a stadium of people cheering made it really easy to understand Neuralink's aims.
My idea of ranting crazy person's website doesn't have facebook share buttons and a popup asking me to subscribe to a newsletter. For reference, here are my ideas of a crazy person's website:
To be fair, waitbutwhy's newsletter isn't spam, it only send about one email per year when new articles are published. As publications are so infrequent, that's a good way not to forget about this blog
Maybe. But the first part with the Jellyfish and the frog and all that evolutionary crap turned me off. Too much of it is written in the Explain Like I'm 5 format.
He is great, but how to put it politely? His enthusiasm for technology and future possibilities blinds him to the reality that many things take longer than what we would wish , and that science requires lot of sluggish, dull work in order to advance.
While this is true, one thing that has changed a lot is how connected all the research labs of the world are across the planet and how that rate is increasing. While it creates a lot of chaos, distraction and trust issues to work through, the scope and scale of collaboration is on the rise which means a lot of work is getting parallelized.
Think of it as a shift from single core to multicore. Dull stuff is going to happen faster than it used too.
It takes time to adjust to hyperconnection. But the benefits will keep showing up and pushing things in the direction of more connection. Look at the number of collaborators on the Black Hole image or gravitational wave detection. You will see things like that increase. And as they increase we learn how to do things better.
At a smaller private event I attended (he was one of three panel participants talking about AI and Tesla), the start was only about 10 minutes late. But it wasn't live streamed. Give it another half hour... :)
Indeed! My friends often refer to his published event times as being in "EST" or "Elon Standard Time", which we consider to be +/- a few half hours of the listed time
Neuralink has a very impressive founding team and I have been periodically checking their homepage for any updates. I am excited to see what they have been working on for all this time.
A key aspect for me will be whether Neuralink can enable brain control interfaces that don't require sending signals to muscles.
A key example that falls short is CRTL-labs' armband. I've seen a few demos that allow you to "control a keyboard with your brain" while wearing the armband. The only problem is that this setup requires you to move your hands as if you were actually typing on a keyboard. If my hands have to move I might as well use a real keyboard. I realize that there are some people that could use CTRL-labs' armband, such as amputees, but it's not compelling to me.
If Neuralink can let me control a computer by visualizing words or something similar instead of physically moving my hand then I'll call it a win.
The CTRL-kit, which is the "real" product they have, uses electromyography, which needs muscle activation to work, so something must twitch at a minimum to send a signal. A good interface may enable controlling multiple things at once, but they would all require muscle activity.
They may have other things in R&D. At least they are closer to having a real product than Neuralink and they started their presentation on time at Re:MARS.
One of my biggest “Wow!” moments of 2018 took place in the offices of neural interface company CTRL-labs. Their demo involves someone playing the old Asteroids computer game without touching a keyboard, using machine learning to interpret the nerve signals that are sent to the hands. But it isn’t quite what you think. Moving things in the digital realm without moving your hands seems startling enough (though it’s worth remembering that it was once considered remarkable to be able to read silently without moving your lips). But that’s just the first stage. Essentially, users of this technology “grow” another virtual hand, which they can move independently of their physical hands. One of the researchers bowled me over when he said he was “working on controlling nine cursors at once.” Gradually, then suddenly, our children will interface with machines in deeper and deeper ways. Humanity is already going cyborg (see trend 1); expect it to accelerate. Don’t fall into the trap of thinking that AI will replace humans when it can be used even more powerfully to augment them.
In the asteriods demo (https://youtu.be/5Z5aZK2C3ew?t=711) the only reason his fingers aren't moving a bunch is because the hand is flat on the table and the table is pushing back. You can see his fingers twitch while he's playing.
This is the biggest overselling point from them. The only people that can grow a "virtual hand" are amputees. Everyone else simply has a shadow hand that follows what their real hand does (or would do if there wasn't a restraint). All of their control interfaces are coupled to what the shadow hand does- i.e., if the shadow hand's index finger curls in, the spaceship spins to the right in the game. In reality it's not a shadow hand they are sensing, but the neurons that control the movement in your hand/arm.
It's a fact of biology that if you are reading nerve impulses from the arm, then to control the computer something must be happening in your arm. Every motor neuron is connected to a muscle. The best we can get is a very sensitive system where your arm doesn't move very much. You cannot (without a million years of evolution) send signals to your arm without it moving. If you were to cut the neurons off from the muscle, you could then send activations without your arm moving. However, you would also lose physical movement ability, and nobody is doing that (except for amputees).
You're right. That said, surgery complicates things. It's a big commitment, it will take time to get approval and it doesn't scale as well for mass production.
I was under the impression that your fact of biology was false.
Specifically that the brain-muscle mapping isn't preprogrammed and the brain learns to control what is there.
For example people with fully formed extra digits can just have a fully functional extra finger.
This does get baked in pretty well but the adult brain can re-learn after a catastrophic injury so it's not hard to believe you could figure out how to add virtual appendages or entirely novel "extremities".
It's not false. You can't send an action potential down a motor neuron connected to a muscle without a reaction from the muscle. Do not confuse the neurons in your brain (which generate new connections all the time) with motor neurons, which is what the CTRL-labs kit is sensing.
The brain does learn to control what is there, and what is there is attached to your muscles. I'll say it again: the CTRL-labs kit works on electromyography, which only works with muscle activity. You will not magically grow new neurons in your arm that are dedicated sole to the control of a computer.
it seems they are targeting motor and sensory cortex. Those do get activated by imagining the task . but it will be a very fuzzy process and require a lot of training. It remains to be seen if 1024 electrodes provide sufficient readings to be useful
As the plain black lettering, "livestream will begin shortly", prosaically displayed across the laptop screen, the laptop's fan whirred away drudging through another hot summer month. The world was about to change, but few expected it. Meanwhile, on the other end of the line, things were unexpectedly delayed.... <cue spooky music>
While we wait for the delayed stream to start, it's worth taking a peek at their job board which has some interesting bits of information in the job descriptions: https://jobs.lever.co/neuralink
For example:
> "As an Optical Engineer at Neuralink you will develop custom optics and imaging systems used directly in our surgical robot, and associated consumables manufacturing systems." [0]
EDIT: I guess the article from The Verge linked in the other comments sheds even more light on that particular bit:
> "[...] the company has developed ”a neurosurgical robot capable of inserting six threads (192 electrodes) per minute [automatically],” according to the white paper. " [1]
Similarly described in the Bloomberg article which was just released [2]
The thing that most excites me about a breaktrough in BCI technology is its possible applications for Brain-to-Brain interfaces a la Asimov (I / We / Gaia) or Ramez Naam (Nexus 5; Very similar to what Neuralink talked about circa 2017)
I've been reading "Carbide Tipped Pens" which is a collection of 17 Hard sci-fi short stories. One of them has a very similar idea. Can't remember which one but I'll post the author if I remember later.
Full speed USB-C allows for high-bandwidth and easy to plug in. It's a pretty good connector if you don't need severe weather resistance.
Using off the shelf hardware saves time for the scientists to develop more meaningful parts of the system.
Using off the shelf hardware makes it seem a little less hacky than if they were running a bunch of jumper wires off a DB-25 connector from an old printer.
One reason I can think of: a 99 cent one off of alibaba might deliver a 100 volt jolt directly to a device that is electrically connected to your brain, killing you instantly.
I'm all for universal standards, but this really does seem like a case where the connection should be a clean sheet (maybe even proprietary) design where safety parameters can be tightly controlled. My guess is that they're using USB during the development phase for ease of development (they probably just haven't spent the time on making a different connector yet). I would be very surprised to see this actually go to market with a USB-C port. Given the notorious issues with USB-C cables, I would be surprised if the FDA even approved a device with a user-accessible USB-C port.
Researchers typically use what's easily available, cheap and well understood. Back in the day it was serial connectors. It's just a bog-standard connector.
A mature product may well use a proprietary connector that isolates against brain-zapping, has better shielding, has no snap-off parts, etc.
>“We will painlessly laser-drill the holes into the skull, place the threads, plug the hole with the sensor, and then you go home,” he says. “It’ll basically be an experience like getting Lasik.”
Maybe I've read a bit too much cyberpunk lore in my day, but this sounds incredibly ambitious. Can't wait for the livestream to start!
As someone who had a magnet inserted and eventually removed from a finger: wait for the 2nd or 3rd revision.
The first versions of the magnet implants had an unnerving tendency to shatter or have their neutral coating breach. And we're talking about the brain here - any damage caused by implanted devices may not be nearly as obvious as an exploded finger. I never miss an opportunity to trot out this quote from Alpha Centauri:
>I think, and my thoughts cross the barrier into the synapses of the machine - just as the good doctor intended. But what I cannot shake, and what hints at things to come, is that thoughts cross back. In my dreams the sensibility of the machine invades the periphery of my consciousness. Dark. Rigid. Cold. Alien. Evolution is at work here, but just what is evolving remains to be seen.
If I think about how omnipresent computers and data are in my life now, I wonder how much we can use that as a starting point for what some of that evolution might look like.
For example...
- I find myself trying to memorize fewer things if I can easily reference them digitally later
- My expectations for having answers at my fingertips has grown such that not having internet or cell service causes anxiety
- My communication is increasingly in a pure visual format, condensing complex ideas, concepts, emotions etc. to a pic or gif. Further, my concept of a photo has bifurcated into more permanent photos I care about, and more ephemeral "pics" that get forgotten about. TBD whether I'd say my overall communication throughout speed has increased
- Phantom phone vibrations when I don't have my device on me
- Shortened attention span due to the wealth of data immediately available, but also many notifications vying for my attention
- A tendency to avoid situations where I'd be without electricity long term
- The ability to enjoy virtual experiences in a way that delivers some of the enjoyment of in-person experiences (think socializing in online games)
Maybe a composite ring with a magnet in it might have a larger market. In the same way ctrl-labs might give you an early-adopter heads-up (although it would be one-way communication)
Simple and painless operation does not mean operation without risks. Even small microsurgery into the brain has infection risk. If successful, Neurallink will be used only for patients in cases where benefits outweigh the risks.
You can start thinking about brain augmentation for non-medical reasons after the technology has been used for decade or so and all the side effects and risks have been minimized.
Their machine is really impressive. I think for a medical device a needle has a great advantage from the safety perspective. You can make a needle that will never penetrate too deep into the brain because it is physically too short. Not so much for a laser capable of burning through skin which, if there is a bug in the controller would burn through the soft tissue in a matter of seconds.
However while it's accepted in regional spoken US English, it's not accepted in written English, even US English.
As for evolution, language does indeed evolve but only when changes are accepted by a significant number of speakers. No matter how much you wish it, you can't simply start forcing a linguistic change on society without society agreeing. If you start to spell "agree" as "agrii" for example, nearly all English speakers will (rightly) tell you that you're wrong, you can't just respond to them with "language evolves". "On accident" is still rare enough that we have a choice to reject it and that's the choice I'm making.
Also, remember that there are non-Americans on this site (I'm one of them). To us, this is simply incorrect. We don't have exposure to this kind of US English in most of the American media we consume (TV, movies etc.).
Symbiosis between man and AI sounds both exciting and incredibly frightening. Imagine on one hand the potential for improving our own intelligence, but on the other hand widening the gap between the rich who can afford this kind of augmentation and the poor who can't.
Sounds like in theory, you should be able to just remove the bluetooth battery unit from behind your ear to completely disable the implants. However, if someone else is in control of your brain, perhaps you wouldn't be able to...
> I'm wondering if someone else could have control of the implant.
My thoughts are that each brain has their own neural network that is unique to the person. This would mean that each neuron, despite similar/same location, will respond uniquely, as our personal neural networks were trained separately (across many years/decades). While I believe it may be possible to train a system to output stimulation to invoke a response, it would be susceptible to the same issues as simple password hashing (that is, while you could break a single hashed password, it would be computationally inefficient). Ignoring quantum computing theory, I think this would then make developing a system that can accurately control someone in LONG sequences would be difficult.
Implant may be individual, but I bet you companies will do their best to turn this into Software as a Service, because of the sweet vendor lock-in, recurring payment money. Consequences, as always, be damned.
Imagine your identity and self being lost to an implant or being around people whose motivations and drive are no longer human.
Imagine the gap between rich and poor to be muddled with an arguable gap between human species.
Imagine the very real arguments about what consciousness is, what a person is with relation to a self spreading to include the machine hardware, perhaps entirely.
>but on the other hand widening the gap between the rich who can afford this kind of augmentation and the poor who can't.
Once two-way communication is achieved, there will be huge leaps in what people can do in their lifetimes. Doubly-so if we figure out how to improve our perception of time and turn ourselves into high speed cameras. We'll have more thoughts per hour, while having access to all of human knowledge and be able to run complicated computer sims (games or actual physical work sims) and do accurate calculations and quickly design stuff with CAD software.
I tried listening in and there were a lot of really intuitively strung-together tangents, each of which could be considered its own separate knowledge domain. I guess I'd say it was very ADHD-friendly, practically an ADHD machine, in that way. I had to give up on the video for now because I kept losing the thread while wondering about this tangent or that one. Will look for a summary later. (I do like that the corresponding gift of Musk's, the connective, big-picture thinking, is being made to work for its money though)
The stitching is interesting - there are lots of near-term implantables with clinical benefits that this could benefit.
I'm less convinced by the overall concept of "I know kung fu", but easy deep brain stimulation or cascade monitoring is probably enough to make a tidy buck.
For all the criticisms of Musk, and recent wackiness, it's moments like these that I thank our lucky stars that there are people like him on this earth. It just seems like these days there are so few people thinking about the long-term future of the human race.
I love it that his solution to the "AI terminator" problem is making a brain-computer interface so that we can have a fighting chance when AI takes off.
I love it that he wants to help us kick our addiction to non-renewable fossil fuels.
I love it that he wants to make us a two planet race so that we don't have all our humanity eggs in one planetary basket.
An electromagnetic pulse, created when a nuclear bomb explodes, that some say will destroy all electronic devices within hundreds of miles that haven't been specifically designed to be EMP resistant.
That's like saying a lions chance against humans is the big teeth. It's dangerous in a particular context, sure, but misses the fundamental assymmetry that took humans from scared or lions to existential threat to lions.
These sorts of criticisms always ignore the prisoner's delima for the sake of expression of moral indignation.
There is no stopping individual agents in a system from doing what helps them most without an authoritarian at the top. Mostly, those authoritarians come with even worse problems so we're left with this imperfect world.
I'd love to see comments on hn not focused on self righteousness and instead realize that there is no one guy at the top that you just have to scream really loud at.
I'm interested in hearing more about about being a charlatan with the meditation app especially considering, as far as I remember, you can get it for free by just asking.
There's two types of people you meet who are into mindfulness: high practicioners (monks) and yoga guy from Los Angeles who is "kinda" into mindfulness but not really
I can only go by what he and other people close to him say but he says he used to do plenty acid and been to retreats in asia for months and months (cumulatively) during his early life and seems to be good pals with people like Joseph Goldstein (who studied under asian teachers in 60s/70s). He has probably experienced all kinds of stuff.
Point being, if you get (at least some of) what there is to get then does it matter where your body was born or what it looks like? Is it a bad thing that western born people are bringing this (buddhist/hindu/jain) thought to the west?
I would revise your statement that there are the monastics who dedicate their lives to this, the lay people who practice and the commoditized 'yoga' as exercise/stretching folk/peddlers who are far removed from its spiritual components.
Eh, if you use the word "mindfulness," you are yoga guy. A monk isn't mindful, he is mortifying his flesh to practice the tenets of the religion he believes in so fiercely enough he is willing to self-imprison to follow it better. What you see as mindfulness is just the surface results of winning that struggle. It is very possible to lose it instead, and monks are often open about the dangers of monastic life.
I think people really don't get religion in this sense. The radical, wild, anarchic aspects of it. Mindfulness is more just a wish for stoicism in religious guise; the idea of being not stoic, and weeping over your prayers in a cell because you feel the weight of the world's sin and know that the time is short will not often occur to people.
>Eh, if you use the word "mindfulness," you are yoga guy
Everything in Buddhism and meditation surrounds around mindfulness/sati/awareness you call it.
>A monk isn't mindful, he is mortifying his flesh to practice the tenets of the religion he believes in so fiercely enough he is willing to self-imprison to follow it better.
Monks have to cultivate the 8 fold path which includes right mindfulness so saying he isn't doesn't make him monk. And also wow that sounds so disrespectful and ignorant.
What exactly is wrong with being a regular person who practices mindfulness? Are the benefits they receive not legitimate in your eyes?
The philosophy I have been exposed to through meditation has helped me better understand how the ego can cause problems. It seems you are rather attached to the idea of a very pure, austere study of meditation and associated philosophies. There are other valid ways of approaching such things that you are unjustifiably disregarding.
Alternatively, you could look at it as someone simply being earlier on their path, and provide encouragement instead of ridicule.
Do you have anything more substantial to say than ad hominem? Like a response to the video I linked instead of grinding your unrelated axe against the guy?
The title of the book: How Science Can Determine Human Values is literally in contradiction with Hume's Guillotine which any philosophy 101 should be aware of.
>Do you have anything more substantial to say than ad hominem?
The reverse argument made here is usually the turkey fallacy. For turkey, all logical evidence points to a continuously improving quality of life, with every need met and a constant availability of food. There’s no evidence that it’s going to be eaten this thanksgiving, so any effort in building turkey-computer neural links is dismissed off hand as being a waste of time.
How does this analogy apply specifically to AI, though? There's also no evidence that God is going to come and pronounce his judgement on us, so any effort in prayer and pious living is often dismissed off hand as being a waste of time. Should non-believers in God reconsider their ways given their knowledge of the turkey fallacy?
Think that’s the premise of Pascal’s wager, that if you simply multiply cost with expectation believing in God is a better bet.
Of course with something like a general AI all bets are off. This neural link think is a horribly bad defense, because of all possible defenses this is the one that could give the AI a direct connection with your brain.
1. paperclip optimizers where a very smart computer you tell to do one menial task like producing as many paperclips as possible or proving a mathematical theorem can turn into a catastrophy as that computer turns all iron on earth into paperclips or into computers that all try to find a solution to the theorem. This also includes computers that we task to "protect" humanity coming to the conclusion that humans having power to kill each other is mankind's biggest threat.
2. crazy would-be dictator who wants to rule over the world and tells an AI to do it or kill all humans or something else.
TLDR: First way: forgetting machines to tell to not kill humans (or not doing it in an effective manner). Second way: some really shit individual explicitly telling machines to kill humans.
The first danger is one we already face: basically since we've had machines there have been accidents with them, also ones involving casualties. In general, the more we care about avoiding casualties the less likely they are. However, it only takes one super intelligent paperclip optimizer to "break lose" so given the high amount of possible casualties, there needs to be a lot of care taken to prevent even one such event.
The second danger needs to be coped as well. One could do two things: very slow deployment of super-AI capabilities at the start, while building AIs that can defend governments and somehow encoding into them how the government works (to prevent parts of the government from using that machine in a coup). The same computers will prevent revolutions though, so I guess we'll see less and less of those. You can think of variations of those ideas like AIs that only enforce asimov's laws or only make sure that we don't use any weapons more powerful than $weapon on each other.
What I don't understand though is how neuralink will help with coping with those threats.
1. Unplug the paperclip optimizer. Blow it up. The problem with the less wrong idea is they keep ascribing more and more godlike powers to AI to counter very common objections to technology. Somehow the entire thing becomes a godzilla like self-sustaining organism that ignores anything we can do or throw at it, and has magical powers. Meanwhile it seems apparently tha major websites can have outages if people go on summer vacation and the interns are on duty.
2. They can do that now. What would an AI do differently that couldn't be accomplished by conventional weapons? How would it do so without using said weapons or any sort of thing that could be done so without it?
The AI thing is just a secular form of the rapture, a particular variant of existential dread for people with little to no religious belief.
> Somehow the entire thing becomes a godzilla like self-sustaining organism that ignores anything we can do or throw at it, and has magical powers. Meanwhile it seems apparently tha major websites can have outages if people go on summer vacation and the interns are on duty.
Sure, the risk is low right now, but the more powerful computers we can build, the larger the potential risk is. Before you manage to press the off button the computer might already have deployed a bioagent or killed countless lives with drones.
> They can do that now. What would an AI do differently that couldn't be accomplished by conventional weapons?
A military made out of humans is subject to human failings. It is generally a big problem that soilders shoot in the general direction of the enemy to not get punishments for not shooting but miss on purpose. As an extreme example, the nazis had to give lots of free alcohol to their soilders so that they'd continue shooting civilians and burying them under new bodies before they have even died. They later invented gas chambers as an easier method to kill masses of people. Compared to humans, an AI is doing what it is being told to. If you tell it "Kill all humans" it will do it.
There are, and have been, many many people working on all of these problems long before Elon Musk (the tech announced today at Neuralink is entirely built on work that was done by people at UCSF and UC Berkeley, and even that is an iteration on technology that was developed by scientists over the past several decades). Neuralink the company was founded by eight people other than Musk. It's a huge disservice to all of these people to give Musk the credit for their work.
Musk sits in a weird position where his unique blend of controversy keeps him in the headlines and ensures he gets linked to these technologies, but that does not mean he is responsible for nor deserves the credit. It could be argued that his "ability" to constantly land in the limelight draws more attention (and thus progress) to these issues, but others would argue that we would be even further along if not for the constant controversy he creates.
One speaker is a professor from UCSF who studied the brain's processing of motor signals. He explicitly credited Musk for having the right vision and the long term planning, and that's why he left his position after 16 years to come work with Neuralink.
No one credits Musk for solving bugs with the software on his products, or creating these brain-computer interfaces. But he can assemble the team to do, and motivate them to keep moving and progressing pretty aggressive schedules. And he frequently gives credit to his team (and doesn't sit there patting himself on the back).
thats a white paper , not the research paper (which is what ppl will read). You can also read it the other way around: He wanted to credit the entire Neuralink team, without claiming to be part if it or leading it
As the other commenters said, that's not a research paper. Here is a link to an actual research paper where (at least some of) the authors work for neuralink.
> He explicitly credited Musk for having the right vision and the long term planning, and that's why he left his position after 16 years to come work with Neuralink.
How do we know that's why, vs his estimation of Musk being the right kind of showman to get a lot of investment.
> How do we know that's why, vs his estimation of Musk being the right kind of showman to get a lot of investment.
Because it is what he explicitly said. As I pointed out in my parent comment.
Of course I'm sure access to capital plays a role. Otherwise it's just someone with a good idea and no money. A meh idea but lots of money also wouldn't attract these kinds of people.
Fred Wilson has a blog post where he outlines the role of a CEO like this:
>A CEO does only three things. Sets the overall vision and strategy of the company and communicates it to all stakeholders. Recruits, hires, and retains the very best talent for the company. Makes sure there is always enough cash in the bank.
Based on everything we saw in the Neuralink livestream, it seems like Elon is nailing all three of these. Doesn't mean he deserves all the credit, but it mean he's doing his job.
... and there were people working on electric cars and rockets before Musk came along too, but somehow he just manages to nudge things along a lot more than the average person!
Technological R&D doesn't get you anywhere on its own. It's an important prerequisite, sure, but just as necessary is the next step, where a company is formed to commercialize/productize novel research through years of schlepping through market education and government safety trials, to pave the way for the technology to become a "safe" product category for other companies to follow on to. There are many technologies stuck between these two stages—thoroughly "researched and developed", but not yet commercialized.
People like Musk (and the people he co-founds these companies with) are important because they're taking nascent product categories that are "stuck" in the R&D stage with little attention being paid to them, and directing large-scale consumer demand onto them in a way that brings profit-driven industry interest—and therefore industry talent—into the picture. Even if it's not Musk's offering that end up winning the space, these efforts redefine the public perception of the category in a way that means that every company in the space wins.
(For another equivalent example: the creator of Bitcoin did more for smart contracts by creating one platform that lead to competitor platforms that actually had smart-contract support, than a thousand academic smart-contract systems projects ever could have.)
> I love it that his solution to the "AI terminator" problem is making a brain-computer interface so that we can have a fighting chance when AI takes off.
This seems like making a well-intentioned medical application that incorporates latest research findings and likely addresses historical downsides of the field (e.g. scarring issues with long term deployment of invasive BCI).
I don't see how that is anywhere close to being related to some sci-fi "AI terminator" scenario though. If you want to go into some cyberpunk fanfic about Musk you can just turn this application around and spin a "AI is now able to fry our brains out" narrative, which is neither helpful, nor realistic. This AI FUD is so weird to me, it's much more likely to be killed by badly written auto pilot for fancy cars, a failed operation to get your brain USB plug, a malicious application of AI by companies or state actors in areas like mass surveillance and population control... than it is for a real, strong AI to suddenly emerge, become sentient and decide that humanity cannot be trusted.
1. This has not been tested on a single human yet, as it has no FDA approval.
2. Preliminary trials in full quadripilegic patients are several away (these are also not yet approved)
3. Should these trials succeed, this will still not be available as an elective procedure for healthy people (that will take much more time)
3. The skull exists and is a hard barrier that is not going away. A decade or so from now, should this be approved as an elective procedure, patients will have to have a hole drilled in their skull (note that most people find LASIK invasive, even after decades of successful surgeries)
4. Patients will also have to become comfortable with thousands of fibers being inserted (albeit in a minimally invasive way) through brain tissue by an automated surgical robot.
5. Should the procedure be successful, patients should finally, at long last, be able to control a mouse, or keyboard, or smartphone using their brain and imagining the movements instead of using their hands.
There is perhaps, a cyberpunk future where crime syndicates mine Bitcoin in the brains of their victims, where malware pipes gigabytes of extremist political memes in seconds through the dorsolateral prefrontal cortex of young adults.
Maybe that will come one day, but this technology is only using the signals generated by the brain to control a mouse and keyboard. This existed twenty years ago in chimpanzee studies. The real innovation here is in materials science and surgery.
This is amazing multi-disciplinary science in the pursuit of advanced medicine, and we should be applauding it for what it is.
So, thank you Elon for funding this -- but more importantly, thanks to all the scientists, researchers, and engineers who have dedicated their lives the advancement of our science and medicine.
I will not be electing to undergo this surgery in the future.
The applications are very very speculative and far-reaching. I think, by the time the applications are feasible there will probably be a way to do minimally invasive craniectomy. The neural implant is impressive work, but anything beyond that is probably going to be very different than is speculated.
> There is perhaps, a cyberpunk future where crime syndicates mine Bitcoin in the brains of their victims, where malware pipes gigabytes of extremist political memes in seconds through the dorsolateral prefrontal cortex of young adults.
Far fetched is an under-statement. It's not possible unless they're planning to do human trials of some non-invasive technology related to the larger product.
This is not a drug, but medical device technology. Therefore, I believe this would qualify for the FDA's pre-market Breakthrough Device Program[1] clearance which, if I'm reading the guidelines correctly, says they can have an expedited decision within 45-60 days from submission[2].
This is under the umbrella of the FDA's Medical Device Safety Action Plan[3] overhaul.
302 comments
[ 5.5 ms ] story [ 283 ms ] threadIt's a long read, but fascinating. The analogy of comparing the brain to a stadium of people cheering made it really easy to understand Neuralink's aims.
http://timecube.2enp.com/
https://web.archive.org/web/20171002152147/http://templeos.o...
https://web.archive.org/web/20180404034756/http://nasimabc.c...
The problem we have is that we have no idea about the language they are speaking.
Edit: IRL Presentation hasn't started either ( https://twitter.com/RebeccaDRobbins/status/11513308672324321... )
A key example that falls short is CRTL-labs' armband. I've seen a few demos that allow you to "control a keyboard with your brain" while wearing the armband. The only problem is that this setup requires you to move your hands as if you were actually typing on a keyboard. If my hands have to move I might as well use a real keyboard. I realize that there are some people that could use CTRL-labs' armband, such as amputees, but it's not compelling to me.
If Neuralink can let me control a computer by visualizing words or something similar instead of physically moving my hand then I'll call it a win.
They also talked about virtual arms, and there was one guy controlling nine cursors simultaneously.
They may have other things in R&D. At least they are closer to having a real product than Neuralink and they started their presentation on time at Re:MARS.
One of my biggest “Wow!” moments of 2018 took place in the offices of neural interface company CTRL-labs. Their demo involves someone playing the old Asteroids computer game without touching a keyboard, using machine learning to interpret the nerve signals that are sent to the hands. But it isn’t quite what you think. Moving things in the digital realm without moving your hands seems startling enough (though it’s worth remembering that it was once considered remarkable to be able to read silently without moving your lips). But that’s just the first stage. Essentially, users of this technology “grow” another virtual hand, which they can move independently of their physical hands. One of the researchers bowled me over when he said he was “working on controlling nine cursors at once.” Gradually, then suddenly, our children will interface with machines in deeper and deeper ways. Humanity is already going cyborg (see trend 1); expect it to accelerate. Don’t fall into the trap of thinking that AI will replace humans when it can be used even more powerfully to augment them.
from https://www.oreilly.com/ideas/gradually-then-suddenly
This is the biggest overselling point from them. The only people that can grow a "virtual hand" are amputees. Everyone else simply has a shadow hand that follows what their real hand does (or would do if there wasn't a restraint). All of their control interfaces are coupled to what the shadow hand does- i.e., if the shadow hand's index finger curls in, the spaceship spins to the right in the game. In reality it's not a shadow hand they are sensing, but the neurons that control the movement in your hand/arm.
It's a fact of biology that if you are reading nerve impulses from the arm, then to control the computer something must be happening in your arm. Every motor neuron is connected to a muscle. The best we can get is a very sensitive system where your arm doesn't move very much. You cannot (without a million years of evolution) send signals to your arm without it moving. If you were to cut the neurons off from the muscle, you could then send activations without your arm moving. However, you would also lose physical movement ability, and nobody is doing that (except for amputees).
Specifically that the brain-muscle mapping isn't preprogrammed and the brain learns to control what is there.
For example people with fully formed extra digits can just have a fully functional extra finger.
This does get baked in pretty well but the adult brain can re-learn after a catastrophic injury so it's not hard to believe you could figure out how to add virtual appendages or entirely novel "extremities".
The brain does learn to control what is there, and what is there is attached to your muscles. I'll say it again: the CTRL-labs kit works on electromyography, which only works with muscle activity. You will not magically grow new neurons in your arm that are dedicated sole to the control of a computer.
For example:
> "As an Optical Engineer at Neuralink you will develop custom optics and imaging systems used directly in our surgical robot, and associated consumables manufacturing systems." [0]
EDIT: I guess the article from The Verge linked in the other comments sheds even more light on that particular bit:
> "[...] the company has developed ”a neurosurgical robot capable of inserting six threads (192 electrodes) per minute [automatically],” according to the white paper. " [1]
Similarly described in the Bloomberg article which was just released [2]
[0] - https://jobs.lever.co/neuralink/c98c011c-0f49-497e-a504-59a5...
[1] - https://www.theverge.com/2019/7/16/20697123/elon-musk-neural...
[2] - https://www.bloomberg.com/news/articles/2019-07-17/elon-musk...
https://www.youtube.com/watch?v=ACYW4RDCS90
https://www.bloomberg.com/news/articles/2019-07-17/elon-musk...
Any speculation around why USB-C was used? Or why this was a detail Neuralink wanted to message on?
Using off the shelf hardware saves time for the scientists to develop more meaningful parts of the system.
Using off the shelf hardware makes it seem a little less hacky than if they were running a bunch of jumper wires off a DB-25 connector from an old printer.
I'm all for universal standards, but this really does seem like a case where the connection should be a clean sheet (maybe even proprietary) design where safety parameters can be tightly controlled. My guess is that they're using USB during the development phase for ease of development (they probably just haven't spent the time on making a different connector yet). I would be very surprised to see this actually go to market with a USB-C port. Given the notorious issues with USB-C cables, I would be surprised if the FDA even approved a device with a user-accessible USB-C port.
A mature product may well use a proprietary connector that isolates against brain-zapping, has better shielding, has no snap-off parts, etc.
given USB C as a "standard" is a complete mess atm...
Maybe I've read a bit too much cyberpunk lore in my day, but this sounds incredibly ambitious. Can't wait for the livestream to start!
The first versions of the magnet implants had an unnerving tendency to shatter or have their neutral coating breach. And we're talking about the brain here - any damage caused by implanted devices may not be nearly as obvious as an exploded finger. I never miss an opportunity to trot out this quote from Alpha Centauri:
>I think, and my thoughts cross the barrier into the synapses of the machine - just as the good doctor intended. But what I cannot shake, and what hints at things to come, is that thoughts cross back. In my dreams the sensibility of the machine invades the periphery of my consciousness. Dark. Rigid. Cold. Alien. Evolution is at work here, but just what is evolving remains to be seen.
For example...
- I find myself trying to memorize fewer things if I can easily reference them digitally later
- My expectations for having answers at my fingertips has grown such that not having internet or cell service causes anxiety
- My communication is increasingly in a pure visual format, condensing complex ideas, concepts, emotions etc. to a pic or gif. Further, my concept of a photo has bifurcated into more permanent photos I care about, and more ephemeral "pics" that get forgotten about. TBD whether I'd say my overall communication throughout speed has increased
- Phantom phone vibrations when I don't have my device on me
- Shortened attention span due to the wealth of data immediately available, but also many notifications vying for my attention
- A tendency to avoid situations where I'd be without electricity long term
- The ability to enjoy virtual experiences in a way that delivers some of the enjoyment of in-person experiences (think socializing in online games)
You can start thinking about brain augmentation for non-medical reasons after the technology has been used for decade or so and all the side effects and risks have been minimized.
https://writingexplained.org/on-accident-or-by-accident
https://books.google.com/ngrams/graph?content=by+accident%2C...
However while it's accepted in regional spoken US English, it's not accepted in written English, even US English.
As for evolution, language does indeed evolve but only when changes are accepted by a significant number of speakers. No matter how much you wish it, you can't simply start forcing a linguistic change on society without society agreeing. If you start to spell "agree" as "agrii" for example, nearly all English speakers will (rightly) tell you that you're wrong, you can't just respond to them with "language evolves". "On accident" is still rare enough that we have a choice to reject it and that's the choice I'm making.
Also, remember that there are non-Americans on this site (I'm one of them). To us, this is simply incorrect. We don't have exposure to this kind of US English in most of the American media we consume (TV, movies etc.).
This is going to be a hell of a ride.
Either that or let some kid tap into my brain. Yeah, sure, that will work out just ... ... ... fine.
You know we're going to pirate and bootleg the hell out of it. You want me to take an ice pick where?
Jesus Christ people are going to be doing this in porn soon.
:)
Involuntary implants for behavior control of violent criminals, or just hackers taking control of regular implants, or worse still ... advertisers
It is not hard to come up with plenty of dystopian scenarios involving this sort of technology.
My thoughts are that each brain has their own neural network that is unique to the person. This would mean that each neuron, despite similar/same location, will respond uniquely, as our personal neural networks were trained separately (across many years/decades). While I believe it may be possible to train a system to output stimulation to invoke a response, it would be susceptible to the same issues as simple password hashing (that is, while you could break a single hashed password, it would be computationally inefficient). Ignoring quantum computing theory, I think this would then make developing a system that can accurately control someone in LONG sequences would be difficult.
Imagine the gap between rich and poor to be muddled with an arguable gap between human species.
Imagine the very real arguments about what consciousness is, what a person is with relation to a self spreading to include the machine hardware, perhaps entirely.
Once two-way communication is achieved, there will be huge leaps in what people can do in their lifetimes. Doubly-so if we figure out how to improve our perception of time and turn ourselves into high speed cameras. We'll have more thoughts per hour, while having access to all of human knowledge and be able to run complicated computer sims (games or actual physical work sims) and do accurate calculations and quickly design stuff with CAD software.
I'm less convinced by the overall concept of "I know kung fu", but easy deep brain stimulation or cascade monitoring is probably enough to make a tidy buck.
I love it that his solution to the "AI terminator" problem is making a brain-computer interface so that we can have a fighting chance when AI takes off.
I love it that he wants to help us kick our addiction to non-renewable fossil fuels.
I love it that he wants to make us a two planet race so that we don't have all our humanity eggs in one planetary basket.
Thank you Elon.
our fighting chance is EMP.
Or if you're no fun, it's an electromagnetic pulse.
there wouldn't be a need for this without rampant, myopic introduction of AI. why not just stop that irresponsible "innovation"?
There is no stopping individual agents in a system from doing what helps them most without an authoritarian at the top. Mostly, those authoritarians come with even worse problems so we're left with this imperfect world.
I'd love to see comments on hn not focused on self righteousness and instead realize that there is no one guy at the top that you just have to scream really loud at.
(1)https://youtube.com/watch?v=wxalrwPNkNI
Point being, if you get (at least some of) what there is to get then does it matter where your body was born or what it looks like? Is it a bad thing that western born people are bringing this (buddhist/hindu/jain) thought to the west?
I would revise your statement that there are the monastics who dedicate their lives to this, the lay people who practice and the commoditized 'yoga' as exercise/stretching folk/peddlers who are far removed from its spiritual components.
I think people really don't get religion in this sense. The radical, wild, anarchic aspects of it. Mindfulness is more just a wish for stoicism in religious guise; the idea of being not stoic, and weeping over your prayers in a cell because you feel the weight of the world's sin and know that the time is short will not often occur to people.
Everything in Buddhism and meditation surrounds around mindfulness/sati/awareness you call it.
>A monk isn't mindful, he is mortifying his flesh to practice the tenets of the religion he believes in so fiercely enough he is willing to self-imprison to follow it better.
Monks have to cultivate the 8 fold path which includes right mindfulness so saying he isn't doesn't make him monk. And also wow that sounds so disrespectful and ignorant.
The philosophy I have been exposed to through meditation has helped me better understand how the ego can cause problems. It seems you are rather attached to the idea of a very pure, austere study of meditation and associated philosophies. There are other valid ways of approaching such things that you are unjustifiably disregarding.
Alternatively, you could look at it as someone simply being earlier on their path, and provide encouragement instead of ridicule.
Completely normal.
Sam Harris is a charlatan for preaching it using "his program": https://www.goodreads.com/book/show/18774981-waking-up
Spare me this book under 4 stars rating.
Ok, good to hear. That's not the impression I got from your comment about the LA yoga guy.
> Sam Harris is a charlatan for preaching it using "his program": https://www.goodreads.com/book/show/18774981-waking-up
What's wrong with the book? I read it and thought it was, on the whole, interesting and useful. Obviously it isn't perfect.
How specifically is Sam a charlatan? What falsehoods does he claim about himself regarding meditation?
Do you have anything more substantial to say than ad hominem? Like a response to the video I linked instead of grinding your unrelated axe against the guy?
>Do you have anything more substantial to say than ad hominem?
Nope, because the title of the book says it all
For an analysis of the state of the field and the surrounding media attention, I highly recommend this blog post by Zachary Lipton [1].
[1] http://approximatelycorrect.com/2017/03/28/the-ai-misinforma...
Of course with something like a general AI all bets are off. This neural link think is a horribly bad defense, because of all possible defenses this is the one that could give the AI a direct connection with your brain.
1. paperclip optimizers where a very smart computer you tell to do one menial task like producing as many paperclips as possible or proving a mathematical theorem can turn into a catastrophy as that computer turns all iron on earth into paperclips or into computers that all try to find a solution to the theorem. This also includes computers that we task to "protect" humanity coming to the conclusion that humans having power to kill each other is mankind's biggest threat.
2. crazy would-be dictator who wants to rule over the world and tells an AI to do it or kill all humans or something else.
TLDR: First way: forgetting machines to tell to not kill humans (or not doing it in an effective manner). Second way: some really shit individual explicitly telling machines to kill humans.
The first danger is one we already face: basically since we've had machines there have been accidents with them, also ones involving casualties. In general, the more we care about avoiding casualties the less likely they are. However, it only takes one super intelligent paperclip optimizer to "break lose" so given the high amount of possible casualties, there needs to be a lot of care taken to prevent even one such event.
The second danger needs to be coped as well. One could do two things: very slow deployment of super-AI capabilities at the start, while building AIs that can defend governments and somehow encoding into them how the government works (to prevent parts of the government from using that machine in a coup). The same computers will prevent revolutions though, so I guess we'll see less and less of those. You can think of variations of those ideas like AIs that only enforce asimov's laws or only make sure that we don't use any weapons more powerful than $weapon on each other.
What I don't understand though is how neuralink will help with coping with those threats.
2. They can do that now. What would an AI do differently that couldn't be accomplished by conventional weapons? How would it do so without using said weapons or any sort of thing that could be done so without it?
The AI thing is just a secular form of the rapture, a particular variant of existential dread for people with little to no religious belief.
Sure, the risk is low right now, but the more powerful computers we can build, the larger the potential risk is. Before you manage to press the off button the computer might already have deployed a bioagent or killed countless lives with drones.
> They can do that now. What would an AI do differently that couldn't be accomplished by conventional weapons?
A military made out of humans is subject to human failings. It is generally a big problem that soilders shoot in the general direction of the enemy to not get punishments for not shooting but miss on purpose. As an extreme example, the nazis had to give lots of free alcohol to their soilders so that they'd continue shooting civilians and burying them under new bodies before they have even died. They later invented gas chambers as an easier method to kill masses of people. Compared to humans, an AI is doing what it is being told to. If you tell it "Kill all humans" it will do it.
Musk sits in a weird position where his unique blend of controversy keeps him in the headlines and ensures he gets linked to these technologies, but that does not mean he is responsible for nor deserves the credit. It could be argued that his "ability" to constantly land in the limelight draws more attention (and thus progress) to these issues, but others would argue that we would be even further along if not for the constant controversy he creates.
I personally have no problem with him.
No one credits Musk for solving bugs with the software on his products, or creating these brain-computer interfaces. But he can assemble the team to do, and motivate them to keep moving and progressing pretty aggressive schedules. And he frequently gives credit to his team (and doesn't sit there patting himself on the back).
Yes, the list of authors on the whitepaper they released is "Elon Musk & Neuralink". I guess his team should be thankful.
https://www.documentcloud.org/documents/6204648-Neuralink-Wh...
https://www.cell.com/neuron/fulltext/S0896-6273(18)30993-0
How do we know that's why, vs his estimation of Musk being the right kind of showman to get a lot of investment.
Because it is what he explicitly said. As I pointed out in my parent comment.
Of course I'm sure access to capital plays a role. Otherwise it's just someone with a good idea and no money. A meh idea but lots of money also wouldn't attract these kinds of people.
>A CEO does only three things. Sets the overall vision and strategy of the company and communicates it to all stakeholders. Recruits, hires, and retains the very best talent for the company. Makes sure there is always enough cash in the bank.
Based on everything we saw in the Neuralink livestream, it seems like Elon is nailing all three of these. Doesn't mean he deserves all the credit, but it mean he's doing his job.
People like Musk (and the people he co-founds these companies with) are important because they're taking nascent product categories that are "stuck" in the R&D stage with little attention being paid to them, and directing large-scale consumer demand onto them in a way that brings profit-driven industry interest—and therefore industry talent—into the picture. Even if it's not Musk's offering that end up winning the space, these efforts redefine the public perception of the category in a way that means that every company in the space wins.
(For another equivalent example: the creator of Bitcoin did more for smart contracts by creating one platform that lead to competitor platforms that actually had smart-contract support, than a thousand academic smart-contract systems projects ever could have.)
This seems like making a well-intentioned medical application that incorporates latest research findings and likely addresses historical downsides of the field (e.g. scarring issues with long term deployment of invasive BCI).
I don't see how that is anywhere close to being related to some sci-fi "AI terminator" scenario though. If you want to go into some cyberpunk fanfic about Musk you can just turn this application around and spin a "AI is now able to fry our brains out" narrative, which is neither helpful, nor realistic. This AI FUD is so weird to me, it's much more likely to be killed by badly written auto pilot for fancy cars, a failed operation to get your brain USB plug, a malicious application of AI by companies or state actors in areas like mass surveillance and population control... than it is for a real, strong AI to suddenly emerge, become sentient and decide that humanity cannot be trusted.
BUT:
1. This has not been tested on a single human yet, as it has no FDA approval.
2. Preliminary trials in full quadripilegic patients are several away (these are also not yet approved)
3. Should these trials succeed, this will still not be available as an elective procedure for healthy people (that will take much more time)
3. The skull exists and is a hard barrier that is not going away. A decade or so from now, should this be approved as an elective procedure, patients will have to have a hole drilled in their skull (note that most people find LASIK invasive, even after decades of successful surgeries)
4. Patients will also have to become comfortable with thousands of fibers being inserted (albeit in a minimally invasive way) through brain tissue by an automated surgical robot.
5. Should the procedure be successful, patients should finally, at long last, be able to control a mouse, or keyboard, or smartphone using their brain and imagining the movements instead of using their hands.
There is perhaps, a cyberpunk future where crime syndicates mine Bitcoin in the brains of their victims, where malware pipes gigabytes of extremist political memes in seconds through the dorsolateral prefrontal cortex of young adults.
Maybe that will come one day, but this technology is only using the signals generated by the brain to control a mouse and keyboard. This existed twenty years ago in chimpanzee studies. The real innovation here is in materials science and surgery.
This is amazing multi-disciplinary science in the pursuit of advanced medicine, and we should be applauding it for what it is.
So, thank you Elon for funding this -- but more importantly, thanks to all the scientists, researchers, and engineers who have dedicated their lives the advancement of our science and medicine.
I will not be electing to undergo this surgery in the future.
Not bad, man. I'd read that book.
This is under the umbrella of the FDA's Medical Device Safety Action Plan[3] overhaul.
[1] https://www.fda.gov/medical-devices/how-study-and-market-you...
[2] https://www.fda.gov/media/108135/download
[3] https://www.fda.gov/about-fda/cdrh-reports/medical-device-sa...