18 – A Popperian Evaluation Of Neuralink’s Presentation

Transcript

This transcript was generated automatically and may contain mistakes.

Dennis: Hey everyone. So just about a week ago, just over a week ago at the time of recording this, Neuralink posted their most recent update. They gave a presentation, first with Elon Musk giving the presentation itself, and then after they had a Q&A where people could send in questions, and I thought it might be interesting to analyze the presentation together. So what I’ll be doing is I will play excerpts of the presentation, probably lots of them, and then comment on them whenever I find something interesting or whenever I think I may potentially have found a mistake.

So this presentation is about Neuralink’s latest hardware developments. And this is where they’re, I think for the first time, I’m not super up to date with everything that Neuralink has presented in the past, but I think this is the first time that they actually present the chip that you can insert into your skull.

And, of course, since I’m interested in epistemology, I’m interested in what Neuralink’s been up to. Although, as you will see, there are, I think, significant differences between epistemology and neuroscience, as I’ve hinted at or explicitly said in the past. And some of the criticism that I’ll present will be in that vein. In any case, so this is the first time, I believe, that Musk presents this chip.

And he will talk us through how the chip is inserted, what it does, and then, like I said, later we’ll get to the Q&A. So let’s jump right in.

Elon Musk: All right, welcome to the Neuralink product demo. I’m really excited to show you what we’ve got. I think it’s going to blow your mind.

Dennis: Spoiler alert, it did not blow my mind, but I think a lot of the stuff that you presented was very cool.

Elon Musk: So the primary purpose of this demo is actually recruiting. So I’m going to emphasize this at the beginning and then again at the end. And we’re not trying to raise money or do anything else. The main purpose of this is to convince great people to come work at Neuralink and help us bring the product to fruition, make it affordable and reliable, such that anyone who wants one can have one. And so I want to emphasize the purpose of Neuralink.

What’s our goal? Our goal is to solve important spine and brain problems with a seamlessly implanted device. So you want to have a device that you can basically put in your head and feel and look totally normal, but it solves some important problem in your brain or spine. And the reality is that almost everyone over time will develop brain and spine problems.

Dennis: This, I think, sounds great. And solving brain and spine problems, that is something I think Neuralink can help with. And I’ll explain later what part of their research I think is promising and what part of their research I think is less promising. But for now, this I think sounds excellent. Helping people with spine and brain problems, that sounds great.

Elon Musk: These range from minor to very severe. But if you live long enough, everyone’s going to basically have some kind of neurological disorder. And these range from memory loss to brain damage. But the thing that’s important to appreciate is that an implantable device can actually solve these problems. I think a lot of people don’t quite realize that. But all of your senses, your sight, hearing, feeling, pain, these are all electrical signals sent by neurons to your brain.

And if you can correct these signals, you can solve everything from memory loss, hearing loss, blindness, paralysis, depression, insomnia, extreme pain, seizures, anxiety, addiction, strokes, brain damage. These can all be solved with an implantable Neuralink. This is an extremely fundamental thing. I think a lot of people don’t quite understand that. The neurons are like wiring. And you kind of need an electronic thing to solve an electronic problem.

Dennis: So here, Musk shows this slide where he says, okay, these are, like he just enumerated, but I just want to enumerate them again just to make it easier to recall them. So he says a Neuralink could help with neurological problems such as memory loss, hearing loss, blindness, paralysis, extreme pain, seizures, strokes, addiction, brain damage, insomnia, depression and anxiety.

Now, for some of these, I think he is correct. I think, for example, you could, if you understand the brain really well and you understand the nervous system in general really well, I think you could cure blindness, say. I mean, we’ve already made a lot of progress in that area. But let’s say, I really don’t know much about the eye or the brain, but let’s say, if there’s, I could imagine that there’s certain types of brain diseases that maybe, you know, deteriorate the optical nerves or whatever it may be, and then a link could maybe help you detect that, or it could maybe even help cure that. So I think that is perfectly plausible, because that strikes me as a hardware issue. I could see the same being true for seizures. I could see the same being true for paralysis. And so in those area, same for hearing loss, in those areas I think Musk is correct that an electrical problem, this is an electrical problem and you can solve it with an electrical solution.

Now, there are areas, though, where he claims that Neuralink could help, where that are neurological problems, such as depression, for example, and anxiety. But I don’t think those are electrical problems. I think those are, those live on a different level of emergence. Now, I’ve, I think I’ve spoken about this stuff in the past on my podcast. So here I’m heavily influenced by David Deutsch’s chapter on abstractions and levels of emergence in The Beginning of Infinity. I think it’s chapter five.

And, basically, I think you will hear me say this a lot throughout this episode, I think Musk is a reductionist, which means that he, whether he realizes this or not, this might not be on purpose, but he basically refuses to think about things on different levels of abstraction, different levels of emergence. So depression and anxiety, those strike me as software issues. They don’t strike me as hardware issues. I could imagine, for example, that if you have a certain brain disease, or so if you have any hardware issue in the brain, that that can also cause software issues up the chain, so to speak. But not every software issue is caused by a hardware issue.

Something else you will notice is that Musk, so first of all, like I said, Musk doesn’t distinguish there either, much, I think, at all really, between the brain’s hardware and the brain’s software. As I’ve said in the past, I think the brain is a computer, and we know from computational universality that whatever the brain does, we could simulate on a MacBook or a PC or whatever other universal computer you have. You don’t have to understand the brain structure or the brain’s hardware in particular or different brain regions or whatever, whatever there may be, because of that computational universality.

David Deutsch once said, I think it was in his interview with CBC Radio, where he said, if you want to understand what was, I think, if you want to understand, this is not an exact quote, but something along the lines of, if you want to understand a word processor or some piece of software that’s running on your computer, then it’s hopeless to study metal and silicon.

If something’s wrong with your word processor, then you should think about maybe there is a bug in the program and you would want to fix the code. You’re not going to, it makes absolutely no sense to take your computer into a repair shop, computer repair shop, if something’s wrong with your word processor. You’d want an explanation on the appropriate level of emergence, and that is the level of software. So I think that is true for depression and anxiety as well.

And what I was going to say earlier is Musk not only seems to try to reduce everything to the hardware level, which unfortunately means that he misses out on the richness of what’s going on in the mind, but he also doesn’t give any explanations of what something like depression and anxiety are. And so, I don’t know, I mean, maybe he has these explanations, he just didn’t, doesn’t give them, but I’m going to go out on a limb and say he doesn’t have them, because I think nobody really does.

So David Deutsch gave a great conjecture, I think, for what happiness is, in The Beginning of Infinity as well. I think that’s chapter, I forget, I think it’s the chapter on bad philosophy. I get, I forget what number that is. Where he says happiness means you’re, you’re able to solve your problems pretty much consistently or continuously. And unhappiness means that you’re chronically balked in your attempts to solve your problems. And I’m not aware of any refutations of that conjecture. So for now, I’m going to, and I think it’s a good explanation. So for now, I’m going to assume that that conjecture is true.

And this could in turn tell us something about depression. So maybe depression is something like, you know, sustained unhappiness or very severe and sustained unhappiness for long periods of time. And maybe such unhappiness can have a cumulative effect. The longer you’re depressed, the more depressed you’re going to be in the future, because you realize that you haven’t solved problems in a long time. And so that compounds that issue for you. So therefore, I’m skeptical that Neuralink could help with depression, for example. And I think anxiety is something very similar. I imagine that there’s something going on in the level of software.

Now with addiction, that one is interesting. I’m not quite sure. That one I could potentially see. And for that, I’m also kind of winging it right now. Depression and anxiety I’ve thought about before. Addiction, I could maybe see is kind of a hybrid issue. I could maybe see that you could have a physical dependency on a substance, say, that makes sense to me. But I could also see that you could have a mental dependency on something.

I wouldn’t consider that a disease of any sort, but you could simply have certain ideas that just, if you act according to those ideas, you then enact certain behavior over and over again. So I’m not sure every addiction has to be physical.

And the same is true for memory loss as well. I can certainly imagine that some types of memory loss are caused by hardware issues. So if you have a disease, a brain disease that eats away at your brain, say, and you’re just literally losing brain matter, and there’s certain ideas stored in that region of the brain, especially if it happens quickly enough, the deterioration, then it wouldn’t surprise me that you then lose those ideas, and so therefore you’ve lost memory.

Although there have been cases, there was one case I’m aware of where somebody had a brain, a fluid buildup in his brain, and I, if I remember correctly, I’m going to put the link in the, in the description, to the article, I think maybe something like three quarters or maybe even more of his brain was completely gone. It was completely replaced with fluid. But because the buildup had been so slow over time, he basically functioned just normal.

He had some tingling or a weakness or something in his leg, in one of his legs, but that was it. He had fine memory, from what I recall, able to perform cognitively, so seemed fine. But that fluid buildup had actually replaced many of the so-called special purpose regions that we today think of as important for, say, the so-called language center of the brain or whatever there may be. It doesn’t seem like these regions are really that important after all, if that account of this patient is true.

But in any case, that’s a whole other can of worms when it comes to computational universality and the allegedly specialized brain regions, which I don’t think it quite works that way either. So again, memory loss, I think, could be a hybrid issue as well. And I could also imagine purely software-induced, or I should be careful with the term induced, I mean memory loss or memory issues that live purely on the level of software.

And what I mean is I can imagine, for example, within the context of the neo-Darwinian theory of the mind, if you’re not familiar with that, listen to my previous episode on that. I think it was two episodes ago where I laid that theory out. The core of the idea there is that the mind is an arena of self-replicating ideas and that memories are just ideas that have been able to stick around and spread through that idea pool for a very long time. So they have either very high fecundity or high longevity.

Actually, for a memory, it always has high longevity. And, and basically, memory loss could, could occur if, let’s say, you have some new ideas, some new ideas evolve in your mind’s idea pool, and they’re so good at spreading through that idea pool that basically what happens is, is they, they, they push away the other ideas that encode those memories. And they, just like in all evolution, they spread through the pool of replicators at the expense of their rivals. So this is really at the core of neo-Darwinian evolution. And so therefore, those other ideas, they die out, so to speak.

I think a lot of memory loss can be explained that way. I think that’s definitely part of the just regular phenomenon of just forgetting something, not the type of forgetting something necessarily where you would consider it a problem. Like if you become so forgetful that it impedes on your ability to live a normal life. I don’t want to call that a disease. I would consider that a kind of software bug or something. So I think memory loss is a hybrid issue.

But when it comes to the things that really are just hardware issues, like say strokes, I would imagine are just hardware issues, hearing loss, like I said, blindness, that makes sense to me, what he said there.

Elon Musk: So current medical research, we’ll just go through what is the state of the art medical research and then what’s the state of the art in what consumers or people in general can get. So the current medical research has shown that you can read neurons in a human’s brain. So there’s something called the Utah Array, which has about 100 channels per array, but it’s kind of like a, it’s a bed of rigid spikes that’s literally inserted with an air hammer. So, you know, that’s, that’s slightly discomforting, I think.

And there’s wires and a box on your head, and so it’s some infection risk. And obviously it will look pretty weird if you’re walking around with boxes on your head. And in order to use it, you have to have an expert medical professional there. And it’s only been done in a few dozen people. But it has served as an important proof of concept that this can be done. So we did want to point this out and show that this actually does work. It’s just not something that the average person could use effectively.

In terms of what is currently available, there is something called deep brain stimulation, where they put a small number of electrodes in your brain and will actually zap your brain with an electric current. And it’s valuable for its uses, but it can’t read or write high bandwidth information. I would say this is sort of a bit like sort of kicking the TV, which does work, but not always.

And it has limitations. Nonetheless, this has greatly helped over 150,000 people. And it’s actually, despite being somewhat of a brute force approach, it has been very effective for a lot of people. And this is what’s currently available. So we want to radically improve this by multiple orders of magnitude, improve it by a factor of 100, then 1,000, then 10,000. So going into the Neuralink architecture…

Dennis: So I’m going to skip this part. I’ll just summarize it real quick. Basically, this is where he shows what the Neuralink looks like, the Link it’s called itself. It’s basically, it looks like it’s about the size of just between a dime and a quarter, maybe, in terms of diameter. And it’s about as thick as maybe four quarters stacked on top of each other. And it has a bunch of metal wiring hanging out of it. And so those wires are then inserted into the top of the brain surface.

And what he explains later is that you can do this very gently and carefully, and it doesn’t cause any bleeding during the procedure. And basically what’s done, he explains how you insert this thing. And that part is really, really cool, I think. So it’s an outpatient procedure. I don’t think they put you under completely.

And basically what they do is they carve out a piece of your skull that’s the size of the Link, and the Link is basically placed right into that hole and replaces, it’s flush, and it replaces your, that piece of the skull. And afterwards the skin, so that, I think they peel back the skin before they place the Link in your skull, and then they put the skin back on top, and you just have a tiny scar, is what Musk claims is how, how it’ll look. So if you have hair, nobody will ever even know that you have a Neuralink. If you’re bald, then there’s just a tiny scar.

But even that, you’d have to look closely. So aesthetically speaking, this is all very promising. And in terms of the procedure, it’s also very promising. I think there’s a doctor present. I might be shaky on this part of the demo. But basically, almost, if not all, of the procedure is actually done by a machine. The machine, I think, is the thing that cuts the hole into your skull, takes out the piece, puts the Link in, actually meshes the metal wiring and the brain together. And all that, I think, is very impressive, very cool engineering that they’re able to do this.

Elon Musk: And it’s also got all the things that you would expect to see, the sensors you’d expect to see in a smartwatch or a phone, like motion measurement, temperature, pressure. So there’s actually a lot of functions that this device could do related to monitoring your health and warning you about a possible heart attack or stroke or other damage, as well as sort of convenience features like playing music.

You can do a lot. It’s sort of like if your phone went in your brain or something. Maybe that’s not a great analogy. All right. So…

Dennis: I just want to comment here real quick on his presentation style. I know this may seem off topic, but I don’t want to limit myself in the criticism that I give. It feels a bit awkward, I’m not going to lie, especially that last remark there. Being a former Apple employee and being a fan of Apple and liking their presentations a lot, I think he could improve his presentation game. I’m not saying he should copy Apple, but I think Apple is great for their, their so-called surprise and delight presentation style. As of late, Apple presentations have gotten a bit more, I don’t know, stilted, and they seem very scripted. And I think there’s way too many people on stage.

Back in the day, when it was just Steve Jobs for like two hours, I think those presentations were great. Very, very good.

This has a very casual vibe, and that’s fine, but it doesn’t seem rehearsed. It doesn’t seem, and not in a good way, it seems a little off the cuff. And so it doesn’t seem to me like he’s speaking very naturally. He’s kind of stumbling over his words, like I am sometimes as I do these podcasts, but my presentations are not as big or important as his. So yeah, I think just overall, as somebody who cares about presentation and the quality and the surprise and delight factor, I think you could do much better there.

Elon Musk: It’s also inductively charged, so it’s charged in the same way that you charge a smartwatch or a phone. And so you can use it all day, charge it at night, and have full functionality. So you would really, it would be completely seamless, and yeah, no wires.

Dennis: I didn’t really get the charging part. I don’t fully understand that. I don’t know if you then, he said you can use it all day and then charge it at night, but you don’t need any cables for it. And that makes sense because, you know, the hole in your head is sealed. But what I don’t understand is, if then, if it’s charged wirelessly, do you then have a little pad or something that you put on the outside of your skull as you sleep? And then that thing is connected to your outlet? I don’t know. I didn’t get that part.

I also wonder why they couldn’t just charge it using, well, why they couldn’t just, I’m sure they’ve thought of it, and I’m sure it’s hard to do, but it would be nice if he could have explained why, for example, couldn’t they just use the brain’s thermal energy or something? Like, if there’s already electricity happening in the brain, right? So, for example, why couldn’t they use that electricity to charge the device, and then it would just be fully charged always, and you wouldn’t have to manually charge it? I think that would have been, even if they can’t do it, it would have been nice to address that. OK, and now skipping some.

Elon Musk: So does it actually work? And what I’m excited to show you, I’ll call it the three little pigs demo. And if our animal handlers, we’re bringing out the pigs. And what we’re going to show you is a, well, I’ll walk right over and show you.

Dennis: So now what Musk is going to do is he’s going to present different pigs at different stages of the procedure, and he’s going to evaluate the results.

Elon Musk: So what we have in pen number one is Joyce, and she does not have an implant. Obviously healthy and happy. All right, we’ll give Gertrude a second, and we’ll move on to Dorothy.

Sometimes the pigs are a little shy. So here’s Dorothy. And in the case of Dorothy, Dorothy used to have an implant, and then we removed the implant. So this is a very important thing to demonstrate, is reversibility. So if you have a Neuralink and then you decide you don’t want it, or you want to get an upgrade and the Neuralink is removed, is it removed in such a way that you are still healthy and happy afterwards? And what Dorothy illustrates is that you can put in the Neuralink, remove it, and be healthy, happy, and indistinguishable from a normal pig.

Dennis: Okay, now there was a lot going on there. So what Musk is saying here, what he’s showing is he’s showing different pigs. There was one pig that hasn’t had the procedure, so this is basically the control pig, as it were. And this is to show what a normal pig looks like, and presumably this pig is healthy and well-fed and whatnot, and what he calls happy. And now we get to the second pig that he just talked about a second ago. And this pig has had the procedure done, but the Link was then removed.

And he says that this is important to show, or this was done to show, that you can go back from having done it. You know, you can decide to get the Link, and then you can remove it again, and that is safe to do. I think this is very important to show that you can do that. You don’t want to marry yourself to this thing for the rest of your life. So I think that was a good thing to try and find out if, you know, a pig would have any health problems, say, after you remove the Link.

Now, he claims that the pig is happy and healthy. And I have some issues with that. So I’m on board with claiming that the pig is healthy, because that we know how to measure. Like, you could, you know, take a blood sample or measure the heart rate or whatever, you know, observe the behaviors. Does the pig do anything weird that it didn’t do before? Those things I think are all pretty easy to explain whether or not the pig is still healthy. Have the feeding habits change, you know, all that kind of stuff.

Now, claiming that the pig is happy, that I think is a whole different ballpark. That’s a whole different game. And the reason is that he’s making a claim about the nature of happiness, and he’s making claims about how a hardware device is impacting a being’s happiness, a sentient being’s happiness, and he’s also implying that pigs are sentient beings.

Now, this is a very unpopular opinion, but I am of the opinion that pigs are not sentient beings. I don’t think any animals are sentient beings. I’m also influenced here by David Deutsch, whose idea is basically that, you know, for everything that we see animals do, we can explain their behavior through the knowledge that is inborn, that they inherited genetically through biological evolution.

And I think that is correct. I have not been able to refute that. And even the most sophisticated animal behavior we see, I think, can be explained in terms of inborn evolution. I’ve introduced this distinction between smarts and intelligence to help clear up the confusion there. So I think there are a lot of animals who are very, very smart, like pigs, for example. Dolphins are usually in that category. Crows are, because they can do sophisticated things. They have sophisticated knowledge within them. That does not mean that they created that knowledge.

That is, so I think, what’s required to consider them intelligent, not just to consider them intelligent, for them to be intelligent and for them to be conscious too. I think these things are related. Or in fact, they’re not just related. I think that, also following David Deutsch, that consciousness is something that is achieved through creativity.

So if you have a creative… Once you run a creative algorithm in an animal or in a person, that entity, that animal or person is then also conscious and has free will and all these other attributes that we associate with intelligence. And I don’t think animals have any of that, unless you somehow implanted some device or something and then ran that algorithm in an animal. And you can run that same program, of course, on a computer, thanks to computational universality. And that, of course, is what the whole project of AGI is about.

But in any case, to get back to what Elon Musk was saying there, he’s making a bunch of different claims, huge claims, but he’s just kind of… this is kind of an off-the-cuff remark about the state of an allegedly sentient being. Now, even granting that pigs are intelligent, so I’m assuming that he’s taking that for granted, and it’s certainly the majority opinion. So, of course, if he were to start talking about, I don’t think pigs are intelligent, animal rights activists would be up in arms and trying to stop Neuralink from doing what they’re doing. So there’s political pressure there that he has to play along if he wants to continue this research. So even if he thinks that animals are not intelligent, although I think he probably does because most people do, he just has to go along with that.

Now, let’s just grant for the sake of argument that pigs are intelligent, that they are sentient. It still wouldn’t follow that the pig is happy, because you cannot know whether anyone, a pig or a person or whatever it might be, is happy unless you’ve explained what happiness is. And just looking at behavior doesn’t cut it. That’s not enough.

So I’ve mentioned earlier David Deutsch’s conjecture that happiness means that you’re continually able to solve your problems. Now, if pigs are intelligent, that means they’re creative. That means they’re problem solvers. So whether or not you’ve had a, so whether or not you’re able to solve your problems, or if the Link has an effect on your ability to solve your problems, that will determine, if David’s conjecture is correct, that will determine whether or not the Link can affect your happiness. But Musk hasn’t made any conjectures about the, the cause of happiness or unhappiness. And so he couldn’t possibly conclude that getting the Link and having the Link removed again has no effect on your happiness.

So this reminds me of explanationless science. I think that’s what Musk is doing here, explanationless science. He’s drawing a bunch of correlations. He sees that a pig who’s never had the Link installed behaves a certain way, and he labels that happy. He labels that as happiness, and he considers the pig’s behavior as a proxy for happiness, because, of course, you can’t directly observe happiness. And if you just have a proxy for happiness, then great, now you don’t even have to explain happiness anymore. You just, you know, put it in quotation marks, say this behavior corresponds to happiness, and now we can measure, quote-unquote, happiness. We can’t necessarily measure its intensity. I don’t think he’s going to make any claims about how happy these pigs are. He’s just saying that they’re happy or they’re not happy. I think in his case, all pigs are happy, so he claims.

So this is also, I think, from the chapter on bad philosophy in The Beginning of Infinity, where you must be careful not to confuse those proxies with the actual thing that you’re trying to explain, or in this case, not even trying to explain, but just claim whether or not it’s present. He claims, he’s labeled these proxies, or this one proxy in this case, happiness. And he says, pig one has never had the procedure. Pig one is happy. We observe the behavior that we label as happiness. That’s our proxy. Although we have no idea what happiness is. And now we also look at pig two, which has had the procedure, and now the procedure was, or now the Link was removed, and the pig behaves in the same way. We can still observe the same proxy being enacted, we observe the same behavior, and so therefore we conclude that that pig is also happy.

That is completely false. It certainly would have been interesting to see that if the, if the pig had suddenly started behaving in weird ways, like I said earlier, that would have been interesting. But none of this is enough to say anything about what happiness is or to make any claims about the inner state of those pigs, if they even have such an inner state. So I think Musk is making several big claims here and frankly has bad methodologies. He’s doing bad science, unfortunately.

Now, you could say, first of all, Musk is not a scientist or a philosopher, right? So whether he’s doing bad science or bad philosophy doesn’t really matter all that much, because he is an engineer. He employs neuroscientists to figure out hardware problems.

Now, I have two answers to that. The first answer is, well, that would be sort of fine if he really confined himself only to working on hardware problems, but he’s not. He wants to solve anxiety and depression and so forth. And like I said, I don’t think those are exclusively hardware problems. I think those are mainly software problems, even if maybe sometimes they’re caused by underlying hardware issues. We should look at them at the appropriate level of emergence.

Second of all, just because Musk is not a philosopher, say, that doesn’t mean that he doesn’t have a certain methodology, that he doesn’t have a certain philosophy of how to solve problems. And like Popper said, all people are philosophers. There’s even a book that’s titled that. And, and that’s because anyone’s methodology, anyone’s way of going about solving problems, that is his philosophy. So it’s impossible not to be a philosopher, in a sense. You might not be a professional philosopher. That, that’s doesn’t matter. You don’t have to be an academic or something to be a philosopher. All people are philosophers.

And I want to share this quote from Karl Popper with you. This is from In Search of a Better World, where he says, quote, all men and all women are philosophers. If they are not conscious of having philosophical problems, they have, at any rate, philosophical prejudices. Most of these are theories which they take for granted. They have absorbed them from their intellectual environment or from tradition.

Since few of these theories are consciously held, they are prejudices in the sense that they are held without critical examination, even though they may be of great importance for the practical actions of people and for their whole life. It is an apology for the existence of professional philosophy that men are needed to examine critically these widespread and influential theories. Theories like these are the insecure starting point of all science and of all philosophy. All philosophy must start from the dubious and often pernicious views of critical common sense. Its aim is to reach enlightened critical common sense, to reach a view nearer to the truth and with a less pernicious influence on human life.

So what Popper is saying here, that all people, even if they’re not consciously aware of having philosophical problems, they at least have certain philosophical prejudices, and they have them simply by virtue of being a person who’s grown up in a certain philosophical environment, and that philosophical environment is the culture in which they grew up. That is my interpretation of what Popper is saying here.

Now, Musk clearly has, he clearly follows a certain philosophical school of thought. He might be aware of this, he might not be aware of. It doesn’t really matter. He very well might be, even though maybe he hasn’t explicitly expressed this in the past. But the philosophical schools of thought that I think he follows are, first, reductionism, because he thinks he can explain software on the level of hardware. That is a classic example of reductionism. And a bit of scientism there, because he thinks that you can scientize, so to speak, philosophical concepts like happiness.

He doesn’t bother explaining these things. He just draws a bunch of correlations, observes pigs’ behavior, and ends up doing explanationless science. And unfortunately, that hinders his progress. So one of the core things I want to get at in this review of his presentation is that philosophy is not just, you know, the navel-gazy, super theoretical, abstract thing. It has real-life implications. And if Musk had a better philosophy, he would make much more rapid and better progress. And therefore, in terms of business, this is important, because he would save time and he would make more money. And surely he’s interested in doing that.

So if he rejected reductionism, and if he tried to create good explanations that are hard to vary, as David Deutsch calls it, I think he could, Neuralink could be making much more progress much more quickly. But anyway, let’s go back to the presentation. So I’m going to skip some, and then he gets to a part where he shows that you can record the neuronal activity and relay that as sound from one of the links in one of the pigs.

Elon Musk: Okay. This is a high energy pig. All right. Gertrude, thanks for coming out. So the beeps you’re hearing are real-time signals from the Neuralink in Gertrude’s head. This Neuralink connects to neurons that are in her snout. So whenever she snuffles around and touches something with her snout, that sends out neural spikes, which are detected here. And so on the screen, you can see each of the spikes from the 1024 electrodes. And then if she snuffles around, touches her snout on the ground, or you kind of feed her some food, pigs love food, then you can see the neurons will fire much more than when you’re not touching the snout. And that’s what’s making the beeping sound.

All right, cool. So as you can see, we have a healthy and happy pig. Initially shy, but obviously high energy and, you know, kind of loving life. And she’s had the implant for two months. So this is a healthy and happy pig with an implant that is two months old and working well. Yeah. All right. Cool.

And then we actually have… I hope this works. So we said, well, what if we do two Neuralink implants? And we’ve been able to do dual Neuralink implants in actually, I think, three pigs at this point, and we have a couple of them here. And we’ve been able to show that you can actually have multiple Neuralinks implanted, and again healthy and happy and indistinguishable from a normal pig. So it’s possible to have multiple Links in your head and have them all be sending out signals and you’re working well. All right, so we just showed you a demonstration of reading brain activity.

Dennis: And now for this part you can’t see this, but he actually explains the visuals pretty well. So just try to imagine in your head what this might look like.

Elon Musk: See? You can probably see that. As I was saying, each of those dots represents a neural spike, and the blue chart at the bottom is showing an accumulation of neural spikes in that region. And in terms of additional brain reading activity, when we have, say, one of our pigs on a treadmill – pig on a treadmill – funny concept, really – and we take the readings from the neurons and we try to predict the position of the joints, and so we have the predicted position of the joints, and then we measure the actual position of the joints, you can see that they’re almost exactly aligned. So we’re able with a wireless neural implant to actually predict the position of all of the limbs in the pig’s body with very high accuracy.

Now, in terms of writing to the brain or stimulating neurons, we also need precise control of the electric field in space and time. We need a wide range of current for different brain regions. Some regions require delicate stimulation, some require a lot of current, and you want obviously no harm to the brain over time. And the way we, part of the way we analyze the stimulating neurons is with a two photon microscopy. I always have trouble pronouncing that, microscopy. And it’s very impressive technology. You can actually literally see in real time how the neurons are firing.

So the red sort of things are the neurons. The red sort of flashing things are the neurons firing. Or I should say the electrodes firing. So the red things are electrodes firing. And then the green are the neuron bodies responding to the current from the electrode. So you can see them lighting up different brain regions. And then by carefully controlling the electric field, you can actually have one electrode influence possibly 1,000 or 10,000 neurons. So although you might only have 1,000 electrodes implanted, you could be influencing millions of neurons. And this is just a similar chart showing stimulation at different power levels.

Dennis: Okay, so now Musk has told us what the device can read and what it can predict off of that data. And now he’s going to say a little more. I’m skipping some here. He’s going to say a little more about how to actually use the device or what it connects to. So it connects to your phone. You’re going to have an app on your phone. And the device has a range of, I think, up to, I think he says 15 feet. I might be wrong about that. And yeah, and then he repeats it. You know, you have wireless charging and whatnot. You can use it all day, that sort of stuff.

Elon Musk: And we’re making good progress towards clinical studies. I’m excited to announce that we received a breakthrough device designation from the FDA in July. Thanks to the hard work of the Neuralink team. So I want to be clear, we’re working closely with the FDA, and we’ll be extremely rigorous. In fact, we will significantly exceed the minimum FDA guidelines for safety. We will make this as safe as possible. Just as with Tesla, while it is legally possible to ship a one-star car, at Tesla, the only cars we make are five stars in every category. So we actually maximize safety, and we’ll take the same approach here at Neuralink.

Dennis: Okay, now we’ve almost reached the end of the presentation. Basically, the last thing that he shows is a slide where he reminds us that the main purpose of the presentation is recruiting, what he really wants to focus on. So he wants to make this appealing to people who might be interested in working at Neuralink. And he shows this slide with nine areas that he wants to hire people for. And those are materials, robotics, software, electronics, chemistry, animal care, biology, surgery, and neuroscience.

And I think what is most glaringly missing from this list is a philosopher. And I mean that, I’m dead serious about that. Even though I know it’s not conventional to hire a philosopher for this kind of thing, like I said, they have a philosophical methodology, and if I may, I think it is a poor philosophical methodology. They have the kind of methodology that will allow them to continue to make quick and great progress in terms of hardware and engineering, but I think they have a kind of methodology that will prevent them from making progress in, or at least significant or rapid progress, areas such as the areas that he wants to work on, like depression, anxiety, making claims about happiness without ever trying to explain what happiness is, that kind of thing.

Now, next they’ll move into Q&A, and there’s a good example of, I think, one of the answers that one of his colleagues gives to the question of consciousness. I’ll get to that a little later. Shows, I think, very clearly why they’re in dire need of a philosopher on the team.

Now, unfortunately, I share Popper’s concern that, unfortunately, most professional philosophers are awful, because they—for them, it’s actually true that most of what they do is just navel-gazing and theorizing without really solving any problems. This kind of armchair philosophy. Whereas the Popperian philosophy, the Deutschian philosophy, that is rapid, active problem solving. It’s just that not every problem can be solved by science.

This is the kind of scientizing, the scientism that I mentioned earlier, where you, you know, you try to label a bunch of things, and, well, that is not unique to scientizing, but, but scientizing is the idea that any problems of any nature, science can solve. And that is not the case. There are philosophical problems that science just has no jurisdiction over. At least at this moment, that might change in the future. For example, once we have a good explanation of what happiness is, once we have a good explanation of what qualia are, maybe we can build a device that we can point at an entity, and then it’ll tell us whether or not that entity is happy. And at that point, it’s a scientific endeavor.

But until then, the philosophy has to come first in this case, and oftentimes it does. So what I would really like to see for Neuralink to make more and better progress and rapid progress is to hire a philosopher. To me, that is the… a Popperian philosopher in particular. To me, that is the glaring omission from this slide of people they want to hire.

Something else I want to mention, in particular when it comes to the criticism I gave of the reductionist approach and this kind of neuroscientific approach of trying to cure mental ailments such as depression and anxiety, I want to clarify that I think neuroscience is a good approach and a fruitful approach for certain things and not for others. I’ve explained some of that. What I think neuroscience is great for is things like brain and spine problems or injuries and that kind of sort. If you have a tumor growing in your eye socket, how is that going to influence your nervous system? That is the kind of thing I think neuroscience is great for. Understanding how the brain works, I think neuroscience is great for that. Trying to help with brain diseases, yes, neuroscience.

When it comes to how the mind works, I don’t see how neuroscience could help us with that. Because again, like I said, those things live on different levels of emergence. For the same reason, again, like I said, that if my browser is broken or my word processor is broken, I don’t open up my computer and start looking at the hardware. Now, if I dump a glass of water on my computer and that leads to software issues, then I might do that. So some hardware issues can result in software issues, but sometimes it’s just a bug in the code. And then looking at the hardware is not going to tell you anything. So that’s where I would make the distinction.

Now, of course you could say, well, Dennis, who are you to say? You’re not a neuroscientist, right? You know nothing about the brain. And to that, I answer, well, first of all, how do you know that? But second of all, it’s okay. And so that’s a bit of an ad hominem argument. So it’s not valid to say that just because somebody is not a neuroscientist, he couldn’t possibly have any criticism to offer for neuroscientists.

I like to think of it as, I like to think of it in terms of this metaphor where, say, you’re at a piano performance, a piano recital or whatever it’s called, and let’s say it’s a master pianist. It’s a great pianist who really knows what he’s doing. And so you’re sitting there. But during the performance, during the piece that he plays marvelously, you notice that at one point he messes up just briefly. He hits the wrong note or something, but you can hear it immediately. Then it’s a little bit like saying, well, you’re not a master pianist, so you don’t know if he did anything wrong. You don’t have to be a master pianist to know that he made a mistake.

Now, some mistakes might be so subtle that only a trained pianist could hear them. I don’t doubt that. But sometimes even a master pianist may make a mistake, that is, glaring enough even to someone who has nothing, who knows nothing about how to play, how to play the piano. And so these are the kinds of things, I think, that apply here too. Even if you’re not a neuroscientist, you can absolutely criticize neuroscientists for their work and their methodology.

And actually, this is where the metaphor kind of breaks down, because the methodology, it’s not only just that they make mistakes, it’s that they have a methodology that prevents them from finding their mistakes, and that’s a, that’s a dangerous kind of error.

So I want to give an example also of something similar. So there is the Redwood Center for Theoretical Neuroscience in Berkeley, but it’s called that because it started in Redwood City on the peninsula in the San Francisco Bay Area. And it was founded originally by Jeff Hawkins, who wrote the book On Intelligence. And I read the book, and from what I remember, he makes several empiricist and reductionist mistakes. And you can see those mistakes coming up again and again in his research. And so also if I look at their website, so let’s see here.

It says here, the Redwood Center for Theoretical Neuroscience is a research center home to faculty, postdocs and students who are working on theories of computation in the brain. That sounds great. And then they say, our goal, quote, the objective of our research is to develop mathematical and computational models of the underlying neurobiological mechanisms involved in perception, cognition, learning, and motor function. We collaborate with experimental neuroscience labs in the design of experiments and in the analysis of neural data. We also train students, blah, blah, blah. Okay.

So basically, I’m going to repeat that first sentence again, because that was really the only important one. The objective of our goal, quote, the objective of our research is to develop mathematical and computational models of the underlying neurobiological mechanisms involved in perception, cognition, learning, and motor function, end quote.

So, I’m not a neuroscientist. I’m, I don’t have a computer science degree. I’m not a mathematician. And yet what these people are, they’re neuroscientists that want to develop mathematical and computational models of the underlying neurobiological mechanisms. I’m also not a biologist, and yet I can tell you, because I know Popperian and Deutschian philosophy, or I know, I know some of it, I can tell you that I can see mistakes in that sentence, and I can tell you that they might learn things about the brain, but I don’t think they’re gonna learn anything about the mind. But they’re saying, for example, that they want to, they want to understand, they want to develop mathematical models of the underlying neurobiological mechanisms of learning.

Now, yes, of course, there’s something neurobiological happening when you’re learning, because you happen to be learning using a computer that’s a brain. But the fact that there’s biology involved is absolutely incidental. Learning is an emergent phenomenon, and it requires creativity. And how creativity works is an epistemological question.

Now, all of these people have impressive degrees, and they’re very, very smart. And so therefore, people who don’t share their degrees are not supposed to criticize their work. But I think that’s wrong. That’s another ad hominem argument. That’s not valid. Like at a piano recital, if you hear a mistake, you can point it out. So yes, it’s the same for the Redwood Center for Theoretical Neuroscience. If they really want to understand how learning works, then that’s an epistemological question. I think I would recommend that they study Popperian epistemology, because Popperian epistemology, without mentioning the brain ever really, except for in his book The Self and Its Brain, I think. But Popper never really wrote that much about the brain. He wrote about the mind, and he was interested in how people solve problems, and you can learn about that entirely without looking into the brain and only looking into the mind.

The same way you can learn how to program, for example, if you want to become a programmer, without knowing all that much about computers. It might be counterintuitive, but it’s true. Now, every now and then, of course, you need to know some basics. You need to know the basics of what memory is and what instruction is, what a processor is, and so forth.

But I’m a programmer, and when I program, the programs that I work on are so abstracted away from the hardware that I almost never care whatsoever about the hardware. I only care about the software level. And I think it’s the same for creativity. It’s a piece of software that’s so abstracted away from the underlying hardware that you don’t have to care about the underlying hardware.

So like I said, learning, that’s an epistemological question. I would not try to bring neuroscience into that. And the same is true of questions such as free will, consciousness. These are all things that I think really have nothing to do with the underlying hardware. So I think what’s great about Popperian and Deutschian philosophy is it gives you a kind of superpower, really, to look at the world and identify people’s philosophical methodologies, even if they don’t explicitly state them, and find mistakes in them. And it offers you a way to detect your own mistakes better as well.

This is one of the reasons I love this philosophy so much, is because it gives me, I like to think, this kind of superpower. And I hope I can get you excited about this as well, because if you learn just a little bit of it, I think you will see a lot more in the world and see mistakes and figure out better ways to, see problems and figure out better ways to solve them.

But in any case, let’s get back to the presentation. So now they get to the Q&A, and what they’re doing is people are writing in questions live, I think through Twitter, and they’re answering some of those questions, and I think some of the questions are prepared as well or something. So we’ll see. I won’t cover all of them, but I’ll pick some interesting ones. So here we go.

Neuralink team member: Garrett asks, what are some of the lower bandwidth activities to target first? Is it muscle movement? Is it auditory signals? What level of bandwidth is required for effective use?

Neuralink engineer: So there’s some low-hanging fruit that I think can really be impactful to help many people’s lives, and that’s restoring movement and communication in, for example, a spinal cord injured patient. And there’s a lot of antecedents in the academic world where there have been very nice demonstrations of doing this, and we think we can take our technology and really bring that to the home, something people can take home with them and improve their lives.

Elon Musk: Fantastic.

Dennis: Yep, I agree that that’s fantastic. Neuralink as a medical device, I think, shows a lot of promise. I think that’s a great area.

Elon Musk: Gil, there’s a fun one next, all you.

Neuralink team member: Yeah, so we’re fielding questions from Twitter, so there’s going to be some funny, funny comments. First question is, who are you and what do you do? I’ll lead with that. And the question is, can the Neuralink chip allow you to summon your Tesla telepathically?

Elon Musk: Definitely, of course. You heard it here first. That’s a definite 100%. Carlos, that is the answer. Just one bit of information. Yeah, it’s very easy. That’s an easy one.

Dennis: So although this question was, I guess, sort of meant as a joke, or maybe it was meant seriously, I don’t know, I was surprised by how confidently Musk and his colleague reacted that they could absolutely do this. Now, I think, I wouldn’t be surprised if this was in theory possible to do this, but from what I’ve seen so far, given an epistemological evaluation of what they’ve done so far, I’m very surprised that they were this confident. Clearly there must be something I’m missing on, not understanding, because I wonder if, you know, it seems to me what they can do so far is they can look at neuronal activity in the brain. They can analyze spikes. They can analyze neuronal patterns.

I don’t know how you would infer anything from that about people’s thoughts, because people’s thoughts are an emergent phenomenon. I don’t think you could simply read off someone’s thoughts from their brain activity. In fact, I think this is the problem of induction again. Imagine you have, let’s simplify it down into something very, very simple. Let’s say you have, you see certain spikes or something in neuronal activity. You see certain patterns, and now you want to match that to what thoughts that person possibly could have thought that could have caused that, those spikes.

Now, this can be reduced to the case where you say you, you, you see something, you observe, observe some phenomenon in nature, and now you want to find the thing that you don’t see, that is the explanation, the cause of that phenomenon. And you never observe that thing. And in fact, there could be a million different, there could be infinitely many explanations that all explain the data that you see. So in their case, there could be infinitely many thoughts the person could have thought that would all result in the same brain activity, I think. But nonetheless, all of those thoughts are different thoughts.

So yeah, this is, like I said, another version of the problem of induction, I think. So I don’t know how they could infer from brain activity what a person is thinking, and I think that they would need to do that to tell that somebody is thinking, please summon my Tesla. So I don’t know. Maybe there’s something I’m missing. If they can somehow figure that out, I think that would be very cool.

Oh, one more thing on this topic. If you have a certain brain pattern in front of you, say, and you want to evaluate it, and you want to see what thoughts that brain pattern might correspond to, what you can do, of course, is guess. You can guess what that would lead to and then try to, and try to test your guess against the data, say, for example, by causing the same brain pattern again and seeing if it corresponds to your, to your theory, if it matches your, your theory. Something like that can do.

But the problem with that is, if you want to guess creatively, well, then the program that runs in the Link would need to be creative, which means it would be a person who’d be intelligent, it’d be conscious, it’d have free will and so forth, because, like I said, I think all these things, they tie together. They’re not separable. So, at that point you would essentially have a second person in your head. It wouldn’t be, that person wouldn’t be part of your brain, but they, they would be in your skull. They would live in that device, and that person would be kind of doomed to evaluate your brain patterns, and a person doesn’t want to do that, at least not forever.

So, and of course, Neuralink would need to crack AGI, because the project of running a person as a program on a computer or, any hardware, like the, the, this device must contain a small computer, that’s the project of AGI. But again, they said it would be super simple. So yeah, I don’t know how they would do it, but it would be very cool.

Neuralink team member: Thank you, Robin. Another question from Twitter. Will you be able to save and replay memories in the future?

Elon Musk: Yes, I think in the future you’ll be able to save and replay memories. I mean, this is obviously sounding increasingly like a Black Mirror episode.

Dennis: I know he was just joking there, but I really wish he hadn’t said that, because there is so much pessimism about technology in our culture and so much awareness of progress in general that I think it’s unfortunate that he’s kind of, I want to say pandering to that pessimism, but he’s giving it a stage in his presentation that should be all about progress and celebrating the technological progress that he’s made.

I’ve enjoyed Black Mirror in the past, and there actually literally was an episode where that was about storing and replaying memories in front of your mind’s eye, and that episode was fun. But what I don’t like about Black Mirror is that they have this kind of tech dystopia theme. That can be fun in fiction, but I wouldn’t take it too seriously. Progress is a good thing.

Elon Musk: But yeah, essentially, if you have a whole brain interface, everything that’s encoded in memory, you could, you could upload. You could basically store your memories as a backup and restore the memories, and ultimately you could potentially download them into a new body or into a robot body.

Dennis: So I think the question was actually aiming at something different. I think the question was about, you know, if you experience something and you, and you have a memory of it, can you store that memory on your Link, and can you then replay it to relive that memory? Whereas I think what, what Musk interpreted the question, I mean, was something like, can you take your, the brain’s entire state and then sort of dump it onto a, you know, hard drive and some other computer? And that you can do, I think. This is a, an idea that has been played with a lot, both in sci-fi and in science, and there is also a project that is related to this effort, is called whole brain emulation. And Musk just spoke of a whole brain interface also.

I want to take a stab at trying to guess what the, what his answer would have been to, or I don’t know if, I’m going to try to guess what his answer would have been to the question the way I understood it. But I think, given Neuralink’s reductionist approach so far, their hardware-driven approach so far, I think they would have a hard time picking out a particular memory and replaying that. Now, if they wanted to do something like that, I think they would need to understand a lot more about the mind than we currently do. It’s not enough to understand the brain very well for those things. I think you need to understand the mind. Again, there’s a crucial distinction there.

The whole brain interface and the, and like taking a dump of the brain and using it as a backup and then running it on a different computer, I think that is doable, and that hopefully will be done very soon. But taking a particular memory and then reliving it, that involves a whole bunch of, mental processes and philosophical ideas that we don’t know yet how they work. Like I said, I’m, I’m, I’m thinking of memory as, or I’m trying to understand it as just an emergent phenomenon of the evolution that occurs inside a mind.

Like I said, I think memories are just ideas. They’re not a particular, like, they’re not a separate type of idea or class of ideas. They have nothing to do with neuronal structures, I think. People usually think of memory as like, or especially associated memories as like, let’s say you go to a certain city and then you meet a certain person there, and so every time you think of that city, you also think of that person. Every time you think of that person, you also think of that city, or most times. So people, I’ve heard these theories where people say, oh, it’s because, you know, the neurons encoding the memory of that city are connected to the neurons, that encode your friend or the person you met there. And I think that’s completely wrong.

You, you just don’t explain those things in terms of hardware. I mean, yes, there’s some wiring underneath there, because it’s the brain, but that’s not the interesting point. The point is that you have certain ideas, and there must be a reason, for why ideas that encode that city in your, in your mind and ideas that encode your acquaintance in your mind, why these ideas relate to each other. They just, they just reference each other. Well, because they happened together. So that just makes sense.

And like I said, I think memories are just, they’re not a separate type of idea. They’re just long-lived ideas. I think that ideas replicate in a mind. I don’t want to go too much into that here. In this episode, I’ve done that before. But memories are just ideas that were able to replicate for a very long time, just like some animals in the biosphere have been around for a very long time, or some genes, I should say. Some genes in the environment have been around for a very long time, whereas other genes, they just don’t make it for very long. They’re not as good at spreading through the population for very long, and it’s the same with ideas in the mind.

We know from Popper that evolution is occurring in a mind. We just don’t know exactly how it looks yet. I hope that the neo-Darwinian theory of the mind sheds some light on that. But if we want to, say, pick out a particular memory and then replay it, re-experience it, relive it, I think that would be, I think we have no clue yet how to do that. Because, first of all, you would need to pick out a particular idea somehow from brain activity, from neuronal patterns. Again, no idea how one would go about that.

And then you would somehow have to induce the qualia of reliving that idea. We don’t know what qualia are yet. So again, there’s philosophy that needs to be done, philosophical work that needs to be done. This is outside the realm of science as of yet. Somebody has to come up with bold explanations of what qualia are, what the mind’s eye is, and then we can start thinking about how one could not only store memories, that strikes me as the easier problem, but relive them, and how to induce through a Neuralink, how to induce this reliving of a memory.

Okay, now a big and also very interesting question here.

Neuralink team member: Awesome. This next question is from ThisIsRex. They’re wondering, can this device be used to explain consciousness?

Elon Musk: In the long term, of course. I can certainly shed some light on consciousness.

Neuralink researcher: This is a really interesting question. I think the answer is yes. And I think one of the reasons that consciousness is so hard is because, like anything in physics, you’re looking at a mapping from X to Y, where X is the neuronal correlates. It’s the thing that’s happening physically. And then Y is this phenomenal state. And historically, we’ve been unable to observe the neuronal correlates very well. And unless it’s in you, we’ve been unable to observe the phenomenal state. So as soon as you were able, neuroscientists are able to personally get these tools where they can see the correlates and they can have the experience, I think the hard problem will vanish very quickly.

Dennis: Okay, so there’s a lot going on there. And I want to quote him again, slowly, because he said a lot of things there. And then I want to deconstruct it bit by bit.

So he said, quote, one of the reasons consciousness is so hard is because, like anything in physics, you’re looking at a mapping from X to Y. Okay, end quote. I want to stop there for a second and say that I find it interesting that he says, like anything in physics. It seems that he, I don’t know who this guy is. Maybe he was introduced earlier and I cut that out, but, he’s one of the people working on, Neuralink, of course. And it seems like he’s saying that the question of consciousness is part of the field of physics. That’s just not true.

I mean, the only relation to physics that I see there is that the kinds of entities that are conscious, namely people, are physical objects, of course, because the information processing has to happen physically. But the question of what consciousness is and how consciousness works, that, it’s not in the realm of physics at all. That’s a philosophical question. So I’m guessing that he also is reductionist, because, this is one of the hallmarks of reductionism, is that you try to reduce any phenomenon whatsoever down to something physical, or trying, trying to make it a branch of physics.

Okay, and then he says, like anything in physics, and then he continues, quote, you’re looking at a mapping from X to Y, where X is the neuronal correlates, something’s happening physically, and then Y is this phenomenal state. And historically, we’ve been unable to observe, blah, blah, blah, end quote.

So here’s another mistake he’s making, and that is he thinks that the question of consciousness will somehow be resolved by looking at mappings. This is in line with what earlier Musk was saying about neuronal states to, say, commanding your Tesla, or what’s it called? Getting your Tesla down, blanking out the word, getting your Tesla to come, summoning your Tesla. And so reading thoughts through some neuronal, through some mapping of neuronal data to thoughts.

And it seems like this colleague is suggesting that the same would be true for consciousness, not only that you could read off, so to speak, a certain thought, but that you could read off an explanation of consciousness. This is, thinking in terms of these mappings is, I think, at least in part a behaviorist mistake. This is characteristic of behaviorism, that instead of trying to explain the underlying phenomenon, you’re only looking at mappings from one label to another.

This is also reminiscent of the labeling of the proxy happiness earlier. You see some behavior and you just label it happiness without really trying to understand what’s actually going on in that black box that you can’t see into. And it’s the same here, where he’s simply assuming, well, if you have these neuronal states, well, then you’re conscious, or that’s part of consciousness, and somehow through these mappings we’ll just be able to read off the explanation. That is just completely false.

The only way to explain anything, including consciousness, is to come up with bold conjectures. You can’t read them off the book of nature. You can’t read them off neuronal structures. It doesn’t work that way.

Okay, and then he says, quote, historically we’ve been unable to observe the correlates very well, end quote. And then, you know, once he says, you know, once neuroscientists have these tools, like Neuralink, to read the neural state, then, quote, the hard problem will vanish very quickly, end quote. So the hard problem is, I’m just reading this off of Wikipedia, is the problem of explaining why and how we have qualia or phenomenal experiences.

So basically, it’s the project of trying to explain what consciousness is, what qualia are, what the mind’s eye is, what it means for something to be like something, to feel like something. This is all part of the hard problem. Yeah, neuronal states are not going to help you with that. So here, in a few simple sentences, we have behaviorism, reductionism, and just bad philosophy and scientism overall, combined. It’s quite impressive, really.

Regarding those mappings and what it could tell us about the mind and consciousness in particular, I want to quote from, this is from Popper’s book All People Are Philosophers, and there, in the second, or in one of the chapters, is called, I’m translating this freely, so in the English edition that might be, might be called differently. It’s called The Second Prelude: The Future Is Open. And it’s a conversation between Popper and Konrad Lorenz, who was an Austrian biologist who became famous for his work on what’s called imprinting, where he did those experiments famously with the geese.

And so they’re having a conversation, and they’re talking about the mind-body problem. So slightly different context, but I think the, this quote still applies, to what we’re in, we’re, we are investigating here. And so Lorenz, says, quote, to say that the soul… And just a quick interjection here, I don’t think he, he’s speaking of some supernatural spiritual thing. I think he means, you know, the self or consciousness. So he says, quote, to say that the soul does not exist, or to say that it’s explicable materially, in terms of matter, is the greatest nonsense that anyone can say. And then he, end quote, and then he quotes his friend Gustav Kramer, who was a German biologist.

And so he quotes his friend, quote, suppose that we, you know, we should have made this utopian success in the research to describe all of the conscious processes into the finest detail, and that we could prove that they correspond to physiological processes bit by bit, then that still would not have solved the mind-body problem. But at most, we could, we’d be justified to say that the psychophysical parallelism was indeed very parallel. So, end quote.

So I think, this also tells us here, even if we were to find that certain, as, as the researcher from Neuralink put, put it, certain mappings from X to Y reliably correspond to, you know, conscious states or conscious processes, that, that just means that, yep, they are indeed very parallel. It doesn’t really tell us anything about what consciousness is or how it works.

To give you an example of what I think of as a more fruitful approach, in his book Objective Knowledge, I think it’s in the appendix, The Bucket and the Searchlight. And I’ve mentioned this before, so if you’ve listened to my podcast, you may be familiar with this already. But Popper mentions, you know, in an attempt to at least kind of put a finger on what consciousness might be, he says, well, consciousness seems to be related to error correction. It seems to be related to discovering that your expectations have been disappointed.

So you can think of this, for example, if you go up a flight of stairs and you reach the end thinking that there is one more step. We’ve all been there. Suddenly you become conscious of the fact that there isn’t an additional step. And you become conscious of the fact that you even had that expectation, because it was disappointed. So Popper says that it is owing to our expectations being disappointed that we even become conscious of having had them. We hold expectations all the time subconsciously, and because they’re met, we’re not really aware that we even have them.

So this is a fruitful approach, I think, a good approach to consciousness, because it’s devoid of anything of the underlying hardware. There’s no reductionism. It’s a straightforward, clear explanation, good methodology. It’s just a bold conjecture. Something like that. Of course, this doesn’t explain what consciousness is, but at least it gives us a hint of when consciousness arises, or what we’re aware of and why. So right there, Popper gave us a very simple and clear and lucid explanation of why we’re conscious of some things and not others. The Neuralink guys, I don’t think, have done that. And if they want to solve the hard problem, they’re not going to be able to read it off of neuronal patterns, like I said. They’re going to have to come up with a bold explanation. And the only way to do that is to conjecture one.

Just to add my own two cents to Popper’s conjecture, something that I think is also true of something that we’re conscious of is, or for why we are conscious of some things and not others, is that all of the thoughts, if you just observe your thoughts popping in and out of consciousness, to use a Sam Harris phrase, you’ll notice that all of those thoughts, I think without exception, I don’t think I can think of any exception, at least from personal experience, all of those thoughts have the appearance of design.

What I mean by that is, this goes back to William Paley with his watch on a heath versus a stone on a heath. A watch, as opposed to a stone, has the appearance of design, because if any of its parts were any different, it wouldn’t perform its function as well.

And I think it’s the same with ideas. We know, thanks to Popper, that the production of the creation of knowledge in our minds arises through an evolutionary process. There’s some variation and selection going on. I think also we should introduce the notion of replication. That is what the neo-Darwinian theory of the mind is all about.

But in any case, the point I’m trying to make is that every thought that we’re constantly aware of already has the appearance of design. And it’s never a garbage idea. But presumably, because there’s mutation involved, like I said, I think ideas self-replicate in the mind. And self-replication is not perfect forever, even if there’s high fidelity. Every now and then mistakes happen.

And sometimes a mistake can probably turn an idea into complete garbage. It can break the structure of the idea. And when that happens, it seems we’re never aware of it. Now, we’re, we’re oftentimes aware of an idea that later we find to be false, say, so we can refute it. But that’s not the same as saying that the idea was garbage. It still had the appearance of design. It was just later falsified or refuted.

But that’s just something to think about. You know, why is it that we’re only ever aware, constantly aware, of ideas that have the appearance of design, when presumably, through mutation that occurs in this mental evolution, there must be many garbage ideas at any point that are completely dysfunctional because of a bad mutation, and yet we’re never aware of them? So it’s explanations like that that I think are fruitful when it comes to furthering our understanding of consciousness.

Okay, and then next, Musk says this in direct response. I didn’t cut anything out in between. Here you go.

Elon Musk: What I find remarkable is that the universe started out as like quarks and leptons. We’ll call it like, you know, hydrogen. And then after a long time, well, what seems like a long time to us, the hydrogen became sentient. It gradually got more complex.

Dennis: So maybe Musk is saying this tongue-in-cheek, at least a little bit, but I want to take him seriously here. I don’t know if he could be any more reductionist. The hydrogen became sentient. No. I mean, first of all, people are made of much more than just hydrogen. There are other elements in our body. Of course, he knows that, so maybe he doesn’t mean that. But second of all, people are software. Like David Deutsch says in The Beginning of Infinity, people, people are, creativity is a property of software.

So invoking elements and the material that we’re made of, that’s not anywhere near what people are or what makes people people. People are problem solvers, and whether they’re made of hydrogen or if they’re made of metal and silicon because they happen to be running on a computer is perfectly incidental. It doesn’t matter. This is another consequence, or this is the consequence of computational universality that I was talking about earlier. And so this kind of reductionism is always going to prevent you from understanding and appreciating the implications of computational universality. And it’s going to make it really hard to understand what software is and what people are.

All right, now skipping some.

Elon Musk: Fantastic. Anyone else? Dan, do you have something to add?

Dan: Yeah, I think one of the things that’s allowed us to make such fast progress in the last few months is the use of pigs as a model. And we started out using, choosing pigs because of very similar anatomy of their skull to humans, same thickness and similar kind of membrane, dural membrane. But then as time went on, we realized that pigs are actually, have amazing other properties. You can train them to walk on treadmills. You can train them to do all kinds of tricks, and also that they have this large representation of the snout in the cortex, which you can very easily stimulate. So the question would be, why are we using pigs? And I think we’ve even surprised ourselves at how useful they are as a model in this respect.

And another important point is that it’s very easy to keep pigs happy. They have very low needs, and so we can build an environment in which they have amazingly good welfare.

Dennis: Okay, there might be some signaling there, how, well, they’re treating the pigs, maybe just to keep animal rights activists at bay, or maybe not. I might be reading that into the situation. However, I do, I was wondering if maybe they chose pigs because they have, you know, quote-unquote, similar neuronal structures to people. I would be very cautious. I definitely want to issue a warning that pigs and people, I mean, hardware-wise, maybe they’re similar, but software-wise, they almost couldn’t be more different.

And so please, Neuralink, do not think that because you keep your pigs happy, or what you call happy, that means that you can infer anything from that about the mental response that people are going to have to the Neuralink. I think people are going to be fine, honestly, with the Neuralink. I don’t think it’s going to lead to any trouble there. It’ll depend on their ideas about the Neuralink. You know, if somebody, all of a sudden, after they get the Neuralink, has deep existential questions about what it means to be a person, and suddenly they’re unhappy because they think they’ve become a cyborg, nothing about their brain has changed. Their ideas have changed. So please don’t read too much into, definitely use the pigs to look at the changes in physical health or lack thereof, but don’t, please don’t infer anything about mental states from pigs and apply that to humans.

Elon Musk: Seem very happy about food. It’s easy to make pigs happy. They love food. Pigs really love food. This is a true thing about pigs. So, you know, that they’re, give them some straw and some things to play with and some friends to hang out with and good food. They’re happy as pigs. Pigs are actually quite similar to people. If we’re going to figure out things for people, then pigs are a good choice.

Dennis: That was the thing I was worried about. I forgot that he was actually going to say that. I didn’t know that he was going to say that right after I said what I just said before. Yeah. I really hope they update their policy on that. Comparing people to pigs is quite dehumanizing, literally and also morally. And again, it’s just, it doesn’t follow. It won’t work. It won’t tell them much about people’s mental states.

All right, next, I’m going to skip a little bit. And then somebody raised the question of telepathy. And I’m going to cut right to where Musk says something interesting. So telepathy being, quoting from Wikipedia, the, quote, purported vicarious transmission of information from one person to another without using any known human sensory channels or physical interaction, end quote.

So the idea is basically, you know, if you have a Neuralink in your skull and another person does too, could those Neuralinks be used to basically download one person’s thoughts directly into the other person’s thought, yeah, the other person’s mind or thoughts or whatever? And that way, would the other person know directly what you’re saying, and, or thinking, rather? Of course, this would be, literally, telepathy, because according to Wikipedia’s definition, that telepathy means you’re not using any sensory channels, and basically Neuralink would then be a sensory channel, just an artificial one. But of course, figuratively speaking, yes, that would be telepathy.

Now, Musk has something to say about this, and then I want to comment on that.

Elon Musk: A lot of our brain thought capacity goes into compressing our thoughts into words, and then you think of the data rate of words. Words are a very low data rate, and we’re putting a tremendous amount of mental energy into compressing the concepts and thoughts in our head into words, and then slowly talking. Speech is so very, very slow. And we could actually send the true thoughts. We could obviously have far better communication, because we can convey the actual concepts, the actual thoughts uncompressed to somebody else.

Dennis: Okay, so Musk is saying that we spend a lot of our mental energy on compressing our thoughts into language, and then he’s saying that it’s, it’s very, it’s a very inefficient way of relaying information to someone, because then they, you know, it’s, it seems to me that he’s saying we spend a lot of effort on it, and then it’s slow, so it’s, it’s, very inefficient. And an inefficient way, language, he’s saying, is an inefficient way of transferring information. And he envisions a way to convey, quote, the true thoughts, end quote, and to, quote, convey the actual thoughts uncompressed, end quote, from one mind to another.

And I think Popper would have had, would have had a thing or two to say about that, and that is that you cannot download a thought from one person’s mind into another. And that is because induction doesn’t work.

I mean, what you could do is you could, if two people can agree on a protocol for how to relay messages using, and how to interpret incoming messages using the Neuralink, that will work perfectly fine. But the act of receiving a thought and understanding its meaning is creative. You can’t just implant that somehow.

I think also this is another area in which the neo-Darwinian theory of the mind can maybe shed some light into why that is. And I think it’s because if you try to, again, if you think of the mind as an arena of self-replicating ideas, that means implanting some idea, throwing it into that pool, is going to drown. And that is because it’s not adapted to spreading in that pool. It has to come from within that pool. Otherwise, it has no chance of surviving.

It’s a little bit like, I take this with a grain of salt, because this analogy doesn’t work exactly, but you can imagine it’s a little bit like taking a species, an animal from one species, it’s really the genes that are the analog here. But imagine taking an animal from one species in one pocket of the biosphere and then putting it in an entirely different pocket of the biosphere, an entirely different environment. It’s not going to survive.

So, yes, you could try to induce an idea. You could try to, you know, ideas are stored somehow in the brain. And, you know, let’s say we crack that, how they’re stored, we know how they’re stored. And you just, you try to download that into another brain. It’s not going to make it, I don’t think. It’s not going to make it in that new pool. It can only make it if it’s adapted to that pool. You can’t do that without knowledge of that pool’s landscape. You can’t do that without knowledge of what the other ideas in the pool are.

So you basically need a full, more or less full understanding of the idea pool in the recipient’s mind in order to successfully download a new copy or upload a new idea into him. And you can’t do that. Also, that would be a gross violation of that person’s privacy.

So this just doesn’t work. But again, what would work is simply agreeing on a protocol that both the sender and the quote-unquote recipient know and understand, and then have some way of interpreting that protocol so that you can creatively understand the meaning of the message. So that would work, and I think that’s still very exciting. It might not be as compressed as Musk would like, but it would still be insanely cool to send someone a message with your mind.

All right, now skipping some more. And by the way, I think I may have forgotten sometimes to announce that I’m skipping some. I don’t think it matters all that much, but keep that in mind if you’re scrubbing the original video and looking for timestamps or something. But now here comes the kicker from Mr. Elon Musk himself.

Elon Musk: I think all these things are great functions for a Neuralink. I think on a species level basis, I think it’s going to be important for us to figure out how we coexist with advanced artificial intelligence. And, you know, I think achieving some kind of AI symbiosis where you have an AI extension of yourself, like a tertiary layer above the limbic system and cortex, and having that symbiosis be good, such that the future of the world is controlled by the combined will of the people of Earth, I think that’s obviously going to be the future that we want, presumably, if it’s the sum of our collective will. And so I think it’s going to be important from an existential threat standpoint to achieve a good AI symbiosis. And that’s what I think might be the most important thing that a device like this achieves.

Dennis: All right, so let’s dig into this. I’m going to be very granular. It might seem like I’m just deconstructing it. But I think it’ll be interesting, because there’s a lot of, just a lot of things, I think, that could be improved that have snuck in here and there. So I’m going to quote him one more time in full, just to get everything into, you know, recent memory more thoroughly. And then I’m going to go over it bit by bit.

So he says, quote, on a species level basis, I think it’s going to be important for us to figure out how we coexist with advanced artificial intelligence. And then skipping some. And achieving some kind of AI symbiosis where you have an AI extension of yourself, like a tertiary layer above the limbic system and cortex. And having that symbiosis be good such that the future of the world is controlled by the combined will of the people of Earth, that’s obviously going to be the future that we want, presumably, if it’s the sum of our collective will. And so I think it’s going to be important from an existential threat standpoint to achieve a good AI symbiosis, and that’s what I think might be the most important thing that a device like this achieves.

So let’s start at the beginning. On a species level basis, I think it’s going to be important for us to figure out how we coexist with advanced artificial intelligence. So that’s the first piece of what he said. Now, as you may recall, going all the way back to episode two, I believe it was, from over a year ago.

So advanced artificial intelligence. Now, Musk is obviously, well, not obviously, but he was influenced by Nick Bostrom’s book Superintelligence. And so I think a lot of what he’s pointing at here is, you know, tip of the hat to Bostrom’s book. It’s not really meant as a tip of the hat. I just mean he’s sort of referencing the concepts from that book. Things like advanced artificial intelligence, and then also superintelligence, although he doesn’t use that word in this quote.

So this idea that you could have advanced intelligence versus regular intelligence, which presumably is us people, and then you could have superintelligence, right? And that is the thing that could think thoughts that we couldn’t even imagine. All of that is refuted by David Deutsch’s idea of the universal explainer. There is no thought that a person could not think. There is no knowledge that a person could not create for himself. So these concepts of advanced artificial intelligence and superintelligence just don’t make any sense there. They don’t work logically. They’re not possible.

So, so if people are universal explainers, so universal explainer just means there’s no problem that can be solved that a person can’t solve. There’s no explanation for how to solve a problem that a person couldn’t come up with. People have that capacity universally. And according to David Deutsch, that is what makes people people.

This is a crucial insight if we want to understand what intelligence is and how creativity works, and why things like advanced, so to speak, or quote-unquote advanced artificial intelligence, and especially superintelligence, can’t exist. And they especially don’t have anything to do with advanced hardware. According to computational universality, once you’ve reached computational universality, the only constraining factors are processing speed and memory capacity. So that is true for universal explainers as well.

Our brains have a certain amount of memory capacity and a certain amount, and here I mean memory as in storage space, not as in remembering things. Our brains have a certain amount of memory storage and have a certain amount of processing speed. Within those constraints, you can think any thought you like. You’re not limited in any way.

It’s very in vogue to think that, you know, we would be to a superintelligence like ants are to us, but that’s completely wrong. Ants are qualitatively different from people, so the argument gets that right, but people are not qualitatively different from any other intelligence. So different levels of sophistication, yes, those exist, and those exist within people too, but that’s not what Musk is referring to here. When he says advanced artificial intelligence, he means something that the rationalist community and the Nick Bostrom influenced, the people that Nick Bostrom has influenced, that refers to a kind of intellect that is not achievable for human beings, for people. And like I said, that can’t exist.

So now, Musk wants to achieve some kind of AI symbiosis, where you have an AI extension of yourself. Okay, if people are universal explainers, then that means you either are a person or you’re not. And it means that you can’t have, because you already have reached the universal repertoire, you can’t have any more of that. You can certainly increase in your sophistication, like some people are smarter than others, for sure. But you can’t have an AI extension of yourself. That, again, doesn’t make sense, unfortunately. You can’t be a person and a half. You’re either a person or you’re not a person.

Now, what you could do is have an extension of yourself that contains software, and you could learn how to use that software. I find that approach very promising. And that could be narrow AI in the sense of, you know, these are machine learning algorithms or something that, you know, automatically and mindlessly process some data for you to look at or investigate. Or, you know, maybe they process sense data so you can have infrared vision, stuff like that. That would be very cool. That would be very useful. But that is just narrow AI, as I call it. That has nothing to do with actual intelligence.

But I think what Musk is actually referring to here is AI as in AGI, as in people intelligence. And so having an AI extension of yourself, that doesn’t work. You’re either a person or you’re not. You can’t have a little bit more of a person in you. If you did have another person in your brain, let’s say, you know, like I mentioned earlier, if you run an AGI on a Neuralink, well, then there is a second person in your head, actually physically located in your head. You won’t share your thought space, but now you would have a second person in your head. And I don’t think that person would want to be in your head.

But for now, it’s, so there’s a moral component there. You don’t want to force an AGI to do stuff. That would be immoral. Just like you don’t want to force any person to do anything. So all the same moral knowledge that we have that applies to people, the same moral principles also apply to AGI, because AGIs are also people. So we wouldn’t want to build a sort of quote-unquote extension of oneself that contains another AGI and then force it, coerce it to, you know, perform mindless tasks that we don’t want to perform. But if that quote-unquote extension just contains, it’s just software, like the same way my laptop on which I’m recording this is a quote-unquote extension of myself, well, then it’s fine. But then it doesn’t really have anything to do with intelligence. It’s just additional software that we can use.

Okay, and then he says, quote, like a tertiary layer above the limbic system and cortex, end quote. Well, like I said before, what the brain hardware looks like doesn’t really matter all that much. The brain’s a universal computer. It must be, because only universal computers have the required repertoire to support the universal explainer. And so the brain is a universal computer, so let’s not worry about different parts of the brain. It’s not important.

Okay. And then he says, quote, and having that symbiosis be good such that the future of the world is controlled by the combined will of the people of Earth, that’s obviously going to be the future that we want, presumably, if it’s the sum of our collective will, end quote. Okay, so he wants the future of the world to be controlled by the combined will of the people of Earth.

Now, there’s a, there’s a few things wrong with this. The first one is that it has actually been mathematically proven. There is a theorem. It was proven by Balinski and Young that there can be no such thing as a combined will of the people, not of Earth or of the United States or of San Francisco or any group of people.

And that might be counterintuitive at first, but they have shown that if you, I think they have five basic assumptions, and they’re all perfectly reasonable, so they’re just things such as, I’m going off of memory here, but there are things like, if every member in the group prefers pizza over hamburgers, this is the example that David gives, I think, in The Beginning of Infinity in chapter 13, if everyone prefers pizza over hamburgers, well, then the entire group should also prefer pizza over hamburgers. That just makes sense, right?

And then you, there’s another rule that says, if one person in the group changes their minds about something else that’s not related to pizza or hamburgers, let’s say they prefer playing basketball over playing golf, that doesn’t change the group’s opinion on whether they want pizza or hamburgers, stuff like that. All these things seem perfectly reasonable, but what Balinski and Young have found is that despite being perfectly reasonable, these axioms lead to inconsistencies and paradoxes, and you cannot derive the will of the people. You cannot derive the will of the group from the individuals. There’s no process of validly doing that.

And so David Deutsch, in chapter 13, called Choices, in The Beginning of Infinity, where he talks about the Balinski and Young theorem, points out that this, this is, is the same as in science, where you cannot validly derive explanations from sense data. You can’t logically, it’s impossible to do. And, and this also rests on Popper’s work on, in political philosophy.

Now, you could argue, well, if societies can’t work this way, you could resort to a sort of nihilism and say, well, then societal planning is doomed, and any decision a society as a whole makes is going to be irrational. But that is not true. The whole point of democracy in the Popperian sense is that you, you try out experiments. You conjecture new experiments that you want to try, and you cast your vote for the experiments that you want to try, and then you cast your vote again next time, next term, to get rid of the experiment.

Now, those experiments are represented by political leaders, for example, or certain policies that you vote for, depending on the political system that you’re in. And that’s perfectly fine. So this is seen as an error-correcting process. And that is all you need to make rational decisions in a society, collectively. You do not need some way of validly deriving, so to speak, collective decisions from individual opinions, from the opinions of the individuals in the group.

Now, what David points out in that same chapter, brilliantly, I think, is that, of course, you could think of a single mind as a group of people, you know, because sometimes we have conflicting preferences. And so you could think of each conflicting preference as an individual in that mind. And you could think of the mind as a whole as the group. And so the same problem applies again to the mind. The same problem of how do you derive the collective will from the individual wills, so to speak, applies to a single mind.

And again, you could resort to nihilism and say, well, we’re hopelessly doomed and people are hopelessly irrational. But of course, there is no such thing as capital R, capital T rational thinking. All you can do is, like Popper said, you can conjecture a solution to a problem and then you criticize it and test it. That is all you need to do, and that is perfectly fine. That’s all there is to rational thinking. Rational thinking does not mean you’re infallible. It does not mean that you’ve found the best solution.

And Popper’s criterion for democracy is simply that you can remove bad leaders and bad policies without bloodshed. Democracy does not mean that you have found a rational way to validly, or you’ve found a rational way to infer what the group opinion should be, or group preference should be, from the individual’s preferences. That is not what that is about at all.

But again, that’s not to mention that this is completely impossible, just mathematically. Of course, I don’t understand the mathematics behind it, but I take David’s and Balinski’s and Young’s word for it that this is, I don’t take a word for it, but I trust for now that that is what our best explanations in math have found, that you cannot derive the will of the people.

Okay, and then he says that’s obviously going to be, this process of derivation, that’s obviously going to be the future that we want, presumably, if it’s the sum of our collective will. So in addition to that being impossible, I also want to say how, the word is going to sound extreme, but I really do mean it, how tyrannical that sounds.

He says, again, I quote, the future of the world is controlled by the combined, or should be in this case, controlled by the combined will of the people of Earth. We don’t want the future of the world controlled by anything. We want the future to be open. We don’t want to have some superhuman, in this case I don’t mean superhuman as in superintelligence, but something above society that manages and controls everything. That sounds like a dictatorship.

What we want are institutions that allow, that sustain, or I should be careful with the word sustain. We want institutions that enable our partly dynamic society to remain dynamic and make progress rapidly and provide stability in times of rapid change. I think that’s roughly how David Deutsch says it. That is what we want. What we don’t want is some, you know, some allegedly infallible decree from some superhuman decision-making agency. That just sounds terrible. That sounds very much dystopian.

Okay. And then he says, quote, and so I think it’s going to be important from an existential threat standpoint to achieve a good AI symbiosis. And that’s what I think might be the most important thing that a device like this achieves.

So, well, here Musk is again influenced by Nick Bostrom and his book Superintelligence, and this idea of existential threat. People like Musk and Bostrom see technological developments as potentially being the end of our civilization. I think that speaks to an interesting conflict in Musk, because Musk is an engineer. He wants progress, but at the same time, he has these ideas about how dangerous technology could potentially be. And of course, what we want is ever more technology as fast as possible. We want as much progress as fast as possible. We don’t need any doomsday scenarios.

And that’s not to mention that something like existential threat cannot possibly follow from AI. I’ve explained that also in episode two, and here I’m again, influenced by David Deutsch. And so a narrow AI cannot possibly be dangerous, because it has a very limited repertoire. If all it does is, say, maximize throughput in a factory, it’s not going to suddenly change its mind and decide to enslave all of humanity, because it simply cannot do that, because it wasn’t programmed to do that.

An AGI is literally a person. Hardware differences aside, it’s qualitatively the same. Now, of course, it could be that an AGI runs on a computer that has way more storage capacity and way more processing power, but that in itself doesn’t make it dangerous. In fact, it allows the AI to do error correction faster, and that includes error correction of its moral knowledge. It includes correcting the errors in its moral knowledge.

So again, we should embrace processing power for our AGI children. That would be a good thing. So if anything, an AGI cannot be any more dangerous than another person. AGI isn’t some transformative technology. There are already people in the world that would not be the first person in the world. So this whole idea about existential threat is, I think, misguided. I think AGI development is perfectly safe, and as I’ve said in the past, I think the real danger lies in hostility toward AGI motivated by ideas like the ones that Musk and Bostrom are proposing.

And I’ve got to say, overall, I don’t mean to be rude to Musk, but I’m getting a bit of a sense that he’s just throwing together a bunch of impressive sounding words. Like, he’s saying stuff like existential threat, limbic system, AI symbiosis, the combined will of the people of Earth. These are all big concepts, and a lot of people are going to be very impressed by that, but they’re not maybe going to be quite able to know what to make of it. But they’re also not going to dare question it, because it’s Elon Musk and you should listen to what he says, or so they think.

So overall, I think this is a reminder that it’s not necessarily good ideas that spread, but simply ideas that are good at spreading. Musk’s ideas spread because he has a platform. And so if he has bad or false ideas, they’re also going to spread despite being false. That is something generally true about meme theory, is your ideas don’t need to be good for them to spread. They just have to be good at spreading.

Okay, back to the Q&A. We’re almost at the end.

Neuralink researcher: I have, in many ways, a very basic science interest, which is I’m really interested in the nature of consciousness. And there’s a lot of very silly philosophy that’s been written about it over the last thousand years. But I think that it’s really, we’ve been very limited by the tools and our ability to interrogate and measure the brain. And as these tools get better, it will pull it into the realm of physics. And it’s really one of the last big, great mysteries in science.

Dennis: So this is the guy from earlier who said that consciousness was a physical, a question of physics. I understand a little bit better now why he thinks that. But I just want to comment that I agree with him that a lot of silly things have been written about consciousness and philosophy over the past thousand years. But it seems to me that therefore he’s rejecting philosophy as a whole. And I think that’s a bad idea. It seems to me that that is his reason for scientizing the question of consciousness. And so that, I’m afraid that’s not going to work.

I also disagree with him about consciousness being one of the last, you know, big questions in science, or whatever he, however he phrased it. There are always going to be infinitely many more interesting problems in science. So we’ll never run out of those.

And so with that, we have pretty much reached the end of the Q&A, and with that, the end of the presentation. I just want to make some closing remarks. The first is I want to get back to that superpower, I called it, that I mentioned earlier, with the piano player. Like I said, I think just even a basic understanding of Popperian and Deutschian philosophy is a bit of a superpower, in that it allows you to find mistakes in your own thinking and in other people’s thinking. And that is not in the sense of, you know, you’re so much better than other people, you’re so much smarter than other people.

It’s just a matter of, we all want to make progress as fast as possible, not all of us, but hopefully as many of us, you know, most of us want to make progress as fast as possible. And that is always going to be a matter of error correction. And the errors that are important to correct in particular are errors that prevent us from detecting ever more errors. And bad philosophy can have that as a consequence, that you’re sort of unable to detect any more errors.

So, like I said, if, you know, if you’re sitting at a piano recital, even if it’s a master piano player, sometimes as a layman you can absolutely hear that the piano player makes a mistake, even though you’re not a master pianist. You don’t have to be a master pianist to detect those mistakes. To fix them, you may need to be a master pianist, but just to detect them, you don’t need to be.

And in fact, once you understand Popperian and Deutschian philosophy a little better, it makes you a better piano listener too. So not only can you detect mistakes, and ever more mistakes, and, you know, the pianist’s performance, but you can also start to tell, just by listening, how he practiced and how he could have practiced better. You know, you can tell about, you can tell the methodology that the pianist used in learning how to play the piano. And again, you can do that too, even if you’re not a pianist.

And from there on out, you can tell what kinds of pieces the pianist could play and what pieces he couldn’t play yet, and why, and how he would need to change his methodology in order to play them. The changes that he needs to make, again, without knowing anything about how to play the piano yourself. And I think this is pretty remarkable.

And like I said at the beginning, I think better philosophy would help Musk and the entire team at Neuralink save time and save money and make more money, because they’re going to make more progress. They’re going to be able to solve more problems with better philosophy. And I think that there are currently some mistakes in their thinking that will prevent them from making as much progress as they could make.

So I like, I like thinking of this presentation that they gave as an example of really how, what a tangible and relevant application of philosophy has in the world today, even in the business world. And I hope that maybe one day Musk has a chance to speak to David, for example. I don’t know if David would want that, but if both feel like it, I think that would be a great exchange. I think Musk would learn a lot.

And I, you know, speaking of, like Neuralink said, they want to hire people. And I definitely think that Neuralink should hire, like, a Popperian philosophy consultant or something to help them criticize their ideas and detect flaws in them and make progress more rapidly. So for the medical applications, they can probably continue to make progress without better philosophy, because that’s mostly low level of emergence. So their reductionism won’t stand in their way. But for any philosophical questions, such as how consciousness works, how to communicate, you know, quote-unquote, telepathically, how the mind works and so forth, for all these questions, those are deep philosophical questions, and they’ll need better philosophy to make progress in those areas.

So to sum up, I wouldn’t necessarily look to Musk for philosophical or political insights, but I would look to him for engineering insights, because I think he’s a great engineer.

All right. Those are my thoughts on Neuralink’s presentation. I hope you enjoyed it. Thank you.

References

This episode makes 2 references to:

• Post ‘The Neo-Darwinian Theory of the Mind’
• Episode ‘16 – The Neo-Darwinian Theory of The Mind’ of this podcast

There is 1 reference to this episode in:

• Post ‘My Thoughts on xAI’