Elon Musk: Neuralink, AI, Autopilot, and the Pale Blue Dot | Artificial Intelligence (AI) Podcast

the following is a conversation with Elon Musk, part two, the second time we talk on the

podcast

with parallels, if not in quality, then an outfit for the great objectively speaking, a sequel to the godfather of all time, part two, like many people know, Elon Musk is a leader. of Tesla SpaceX, his link and the boring company, perhaps less known is that he is a world-class engineer and designer who constantly emphasizes first principles to tackle big engineering problems that many before him will consider impossible like the scientists and engineers who the most of us consider. We don't question the way things are done, we simply follow the momentum of the multitude of revolutionary ideas that change the world on small and large scales.

When you go back to basics and wonder if there is a better way, this conversation focuses on the incredible engineering and innovation being done in brain-computer interfaces and neural linking, this work promises to help treat neurobiological diseases to help us better understand the connection between the individual neuron and high-level function of the human brain and eventually one day expand the capacity of the brain through two-way communication with computing devices, the Internet and

artificial

intelligence

systems, this is the

artificial

intelligence

podcast

, if you enjoy it , subscribe on YouTube, Apple Podcasts, Patreon-supported Spotify, or just connect with me on Twitter.

Alex Friedman spelled Fri D ma M and now as An anonymous YouTube commenter refers to our previous conversation as the quote. First historic video of two robots chatting without supervision. Here is the second time. The second conversation with Elon Musk. Let's start with an easy question about consciousness. In his opinion, consciousness is something unique to humans. something that permeates all matter almost like a fundamental force of physics. I do not believe that consciousness permeates all matter. I just think yes, there is a philosophical message, how would you say it's true? That's a good point. I believe in the scientific message.

Don't ruin yourself. Don't worry about anything other than the scientific method, it's like if you can't test the hypothesis then you can't come to a meaningful conclusion that it's true. Do you believe that understanding consciousness is within the reach of the science of the scientific method? We can dramatically improve our understanding of consciousness. It's hard to say we understand something with complete precision, but can we dramatically improve ours? I think the answer is yes. This Nai system, in his opinion, I have to have consciousness to achieve intelligence at a human or superhuman level. you need to have some of these human qualities that consciousness can be a body it can be a fear of mortality ability to love those kinds of silly human things it's different you know there's this scientific method that I really believe in where something is true to the degree to which it's testable and otherwise you're really just talking about known preferences or totally untestable beliefs or you know that kind of thing, so it ends up being kind of a semantic question where we were conflating a lot of things with the word intelligence, if we analyze them and say that you know everything, we are heading towards the future where an AI will be able to think better than us in every way, then the answer is unequivocally yes, so that an AI system that needs to think better than us . we in every way also need to have the ability to be self-aware and Anjali will be self-aware yes, that is different from consciousness.

I need to be in terms of what consciousness feels like consciousness is in a different dimension. but this could just be an illusion, you know, if you damage your brain in some way physically, you damage your consciousness, which implies that consciousness is a physical phenomenon and, in my opinion, the point is that, I think they are really quite It is very likely that digital intelligence will be able to surpass us in every way and will soon be able to simulate what we consider consciousness, so that we will not be able to tell the difference, and from a scientific point of view. method, it could also be consciousness if we can simulate it perfectly if you can't tell the difference and this is a kind of Turing test, but think about a more advanced version of the Turing test if you're talking to a digital super intelligence and you can't tell If it's a computer or a human, let's say just having a conversation on the phone or a video conference or something where you think you're talking, look, look. as a person you make all the right inflections and movements and all the little subtleties that make up a human being and you speak as a human you make mistakes like you're listening look at that and you literally can't tell if you're actually conversing with this a person or an AI could also Being human, so on a darker topic, has expressed serious concern about the existential threats of AI, is perhaps one of the biggest challenges facing our civilization, but as I would say we are sort of like optimistic descendants. of apes, perhaps we can find several paths to escape the harm of AI, so if I can give you three options, perhaps you can comment on which one you think is the most promising, so that efforts on ape safety can be intensified.

AI and beneficial research and benefits are carried out. hoping to find an algorithmic or perhaps political solution to becoming a multiplanetary species as quickly as possible and three is merging with AI and riding the wave of that increasing intelligence as it continually improves what it believes is more promising and more interesting as a civilization we should invest in, I think a lot is being done in response to investment, while the lack of investment is in AI safety and, in my opinion, there should be a cup, an agency that oversees everything related to AI to confirm that it does not pose a risk to public safety, just as there is a regulatory authority for this, such as the Food and Drug Administration, which is four-cornered automotive safety , there is the FA for aircraft safety, which generally concludes that it is important to have a government arbitrator or an arbitrator who serves the public interest by ensuring things are safe when there is a potential danger to the public.

I would say that AI is unequivocally something that has the potential to be dangerous to the public and therefore should have a regulatory agency just like other things that are dangerous to the public have a regulatory agency, but let me tell you that the problem with this is that the government was very slow and the speed with which a regulatory agency is usually created is that something terrible happens, there is a huge public outcry and years after that, a regulatory agency is implemented or a rule is implemented that requires something like seat belts, it was already known for a decade or more that seat belts would have a massive impact on safety and save so many lives in serious injuries and the auto industry fought against Requirements put seat belts with nails and teeth, that's crazy, yeah, and we honor hundreds of thousands of people who probably died because of it and said people wouldn't buy cars if their seat belts were obviously absurd, you know, or look at the tobacco industry and how long.

They fought against anything related to smoking, that's part of why I helped make that movie. Thanks for smoking. You can see how pernicious it can be when you have these companies that effectively achieve the government regulatory capture that the bad people in the AG community refer to. to the advent of digital superintelligence as a singularity that does not mean that it is good or bad, but that it is very difficult to predict what will happen after that point and that there is a certain probability that it will be bad, and that it will probably be bad. will be good or if they want you to affect that probability and make it more good than bad, let me talk about the question of the fusion with AI and the incredible work that is being done in your link.

There are many fascinating innovations here in different disciplines. going through that, the flexible cables, the robotic sewing machine that responds to the movement of the brain, everything around it ensures safety, etc., so we currently understand very little about the human brain. Do you also hope that the work on the neural link will help us understand more about our the human mind about the brain, yes, the work in your style will definitely give a lot of information about how the brain, the mind works at this time, just the The data we have on how the brain works is very limited, we have collected functional MRI, which is what it is.

That's like putting, you know, a stethoscope on the outside of a factory wall and then putting it all over the factory wall and you can hear the sounds, but you don't know what the machines are actually doing, yeah, it's hard. . You can infer some things, but it's a very poor brushstroke. To really know what's going on in the brain, you really need to have high precision sensors and then you want to have stimuli and responses, like if you activate a new one, how? you feel, what you see, how it changes your perception of the world that you are talking to physically, just getting closer to the brain, being able to measure signals in the brain, yes, it will allow us to open the door and enter the factory, yes, being able to having high precision sensors that tell you what individual neurons are doing and then being able to activate a neuron and see what the responses are in the brain so you can see the consequences of if you activate this neuron, what happens, how it feels, what it is It will be a really profound change to have this in people because people can articulate their change as if there is a change in mood or if they have to know if they can tell you if they can see better, hear better or be able to train. better or worse phrases or you know their memories are activated or that kind of thing, so on the human side there is this incredible general malleability, plasticity of the human brain, the human brain adapts, adjusts, etc., so it's not so plastic to be completely frank. there is a firm structure but yet there is some plasticity and the open question is, if I could ask a broad question, is to what extent can that plasticity be used on the human side, is there some plasticity in the human brain and in the machine side we have neural networks machine learning artificial intelligence is able to adjust and discover signals so there is a mysterious language that we don't fully understand that is inside the human brain and then we try to understand that language to communicate both directions so that the The brain adjusts a little, we don't know how much, and the machine is adjusting.

Where do you see them trying to unite, almost like an alien species? Try to find a communication protocol that works. Where do you see the biggest one? The biggest benefit comes from the machine side or the human side. Do you see them both working together? I think the machine side is much more malleable and the biological side is very large, so it will be the machine that adapts to the brain. The only thing that is possible is that the brain can adapt so well to the machine. You can't make neurons start considering an electrode as a corner.

Another neuron because you don't just dislike the pulse and then something else is pulsing. Look, so there's this. that there is that elasticity inside that we think is something that can happen, but the vast majority of the malleability will have to be on the machine side, but it's interesting when you look at that synaptic plasticity in the interface ID, there could be as an emergent plasticity because it is completely different, it is not like in the brain, it is a completely different extension of the brain. You know, we may have to redefine what it means to be malleable for the brain, so maybe the brain can adapt to external interfaces.

There will be some adjustment in the brain because there will be something that reads and simulates the brain and then it will adjust to that thing, but well, if the vast majority of the adjustment will be on the machine side, this is only if this is what it has. otherwise it ultimately won't work, since we don't currently operate in two layers, we have a kind of lamech, like the main primitive brain layer, which is where all our types of impulses come from, it's kind of like we have we have like a monkey brain with a computer attached to it, that's the human brain and many of our impulses and everything are driven by the monkey brain and the cortex computer is constantly trying to make the monkey brain happy Montek.

It's not the cortex that runs the monkey, it's the monkey's brain that runs the cortex, you know, so the cortex is the part that tells the story of everything, so we convince ourselves that it's more interesting than just the monkey's brain. Monkey, the bark is like what we do. You'll call human intelligence, you know, it's like an advanced computer relative to other creatures, like other creatures don't have it, or we don't really, they don't have the computer or they have a very weak computer inrelationship with humans. but it's like it seems like surely the really smart should control the dumb, but I actually don't think this works, so do you think about some of the same kind of machine learning methods, whether it's natural language processing?

Applications for communication between the machine and the brain will be applied to learn how to do certain things such as body movement, how to process visual stimuli, etc. Do you see the value of using machine learning to understand the language of two-way communication with the brain, yes, absolutely, maybe we are a neural network and, you know, AI is basicallyIt is known that it is like the digital neural network interacting with the biological neural network and hopefully take us for a ride, yes, but the vast majority of those that are not of our intelligence will be digital, there are no such things as the difference in intelligence between Your cortex and the limbic system are gigantic Your living system doesn't really understand what the heck the cortex is doing.

He's just literally hungry, you know, or you're tired, or angry, or he's sexy, or something like that. You know, it's just an announcement and then it tells you, that's all. impulse to the bark and Tails to the bark to go satisfy that then a huge amount of thought like truly this stupendous amount of thought has been dedicated to sex without purpose, without provocation, without procreation, which is actually a pretty stupid action in the absence of procreation it's a little silly the reason you do it is because it makes the limbic system happy that's why but it's pretty absurd really well, the whole existence is pretty absurd in some sense, yeah, but I mean, This does a lot of calculations on how I can do more of that without co-creation even being a factor.

I think it is a very important area of ​​research for NSFW, an agency that should receive a lot of funding, especially after this decision, whether to propose the formation of a new agency, oh boy, what is the most exciting thing or some of the most What exciting things do you see in the future? Impact of the neural link on both the size and engineering of a broad social impact. So in your link I think that will first resolve. many diseases related to the brain, so they arise like autism, schizophrenia, memory loss, as everyone experiences memory loss that at a certain point of age, parents cannot remember the names of their children and that kind of thing, so that neural link can do a lot of good.

To resolve critical damage to the brain or spinal cord, there's a lot that can be done to improve people's quality of life and those will be those three steps along the way and then ultimately you're looking to address the risk of existential risks. associated with digital super intelligence, like we don't feel smarter than a digital supercomputer, so if you can't beat them, join them and freed, we won't have that option, so you're hoping your link can be a kind of connection to allow us to merge to ride the wave of the improvement of AI systems. I think the chances above zero percent are non-zero yes there is a chance and that's what I've seen getting sillier yes so I say there's a chance he's saying one in a billion or one in a million, whatever, dumber and sillier, you know, it went from maybe one in a million to improving, maybe it will be one in a thousand and then 100, then one in ten, it depends on the rate of improvement. of the neural link and how quickly we can progress, you know well, I've talked to some people here, pretty brilliant engineers, some I'm excited about, yeah, I think it's fundamentally good, you know who you know, to give someone all their money back . motor control after having a spinal cord injury, you know, restoring brain functionality after a stroke, resolving debilitating genetically orange brain diseases, I think all of this is incredibly cool, and to be able to do that, you have to be able to interact with neurons in detail. level and each build fire are not write neurons, read the write neurons and then effectively you can create a circuit, replace what is broken with silicon and actually fill in the missing functionality to them and then with Over time, we can develop a tertiary layer.

So if the limbic system is a primary layer, then the cortex is like a sector, the second layer now and I said you know the cortex is a lot smarter than the limbic system, but people generally like the fact that they have a living system and a cortex I have met someone who wants to guide either of them there like a girl keeping both that's great the limbic system is a fun thing tell us what the absolute fun is and then you guys usually don't lose the cortex either , OK? They are like having the cortex and the limbic system, yes, and then there is a tertiary layer that will be the digital super intelligence and I think there is room for optimism given that the cortex, the cortex, is very intelligent and the limbic system is not. , and they still work. together, well, maybe they can be a tertiary layer where the digital super intelligence is located and that will be much smarter than the cortex, but will still coexist peacefully and in the end in an EIN way with the cortex and the limbic system, which is a super exciting future both in The lowest engineering I saw being done here and the real possibility in the coming decades it is important that Norling solve this problem sooner rather than later because the point at which we have digital superintelligence is when we pass the singularity and Things become very uncertain, it doesn't mean they are necessarily bad or good by the point we pass the singularity.

Things become extremely unstable, so we want to have a human brain interface before the singularity or at least not long after to minimize it. existential risk to humanity and consciousness as we know it, but there is a lot of fascinating real engineering and low level problems here in your link that yes, quite exciting, what are the problems we face in your art, materials science, electrical engineering , software, mechanical engineering, micro. Manufacturing is a set of engineering disciplines, essentially, that's where it comes down to you have to have a small electrode, so small that it doesn't damage the neurons, but it has to last as long as a person, so it will last for decades and then. you have to take that signal, you have to process that single signal locally at low power, so we need a lot of chip design engineers you know, because we're going to process the signal and do it very energy efficiently. way so that we don't heat up your brain because the brain is very sensitive to heat and then we are going to take those signals.

I'm going to do something with them and then we better stimulate interest in stimulating the back as well. you know, so you could buy directional communication, so good in materials science, software, mechanical engineering, electrical engineering, travel design, microfabrication, that's what we need to work on, we need good materials science so we can having tiny electrodes that last a long time and, like the difficult thing about scientific problems is difficult because you're trying to read and simulate electrically in an electrically active area, your brain is very electrically active in electrochemically active, so how do you have a layer? the electrode that doesn't dissolve over time and is safe in the brain, this is a very difficult problem and from time to time, how do you collect those signals in the most efficient way because you really only have very small amounts of energy to process those signals, you know, and then we need to automate everything to make it like LASIK, you know, so it's just not like, if this is done by neurosurgeons, there's no way it can scale to large numbers of people and it needs to scale to large numbers. of people because I think ultimately we want the repeated future to be determined by a large number of humans.

Do you think this has the possibility of revolutionizing the surgery period? So neurosurgery and Ellis, yeah, sure, he must be as vague as if you met him. LASIK had to be done by hand, not done by hand by a person, that wouldn't be cool, you know, it's done by a robot and they'll take you out of the mall, they just need to make sure you head to my position and then they just press the button and off they go, it's a clever summons and soon Auto Park is faced with the beautiful mess of parking lots and their human non-verbal communication.

I think it actually has the potential to have a profound impact on changing the way our civilization views AI. in robotics because this is the first time human beings, people who don't own them or test them. Eve never saw it, it doesn't hurt that a Tesla can see hundreds of thousands of self-driving cars, yeah, do you see it this way, almost like? an educational tool for the world about AI, do you feel the burden of that, the excitement of that or do you just think it's a smart parking feature? I think you're getting at something important, which is that most people have never really seen a robot or and what's the card that's autonomous is a four-wheeled robot yeah, it communicates a certain type of message with everything from safety to the possibility of what AI could bring its current limitations its current challenges is what is possible do you feel the burden that almost as an educator communicator to the world about AI, we were actually trying to make the lives of the fuels would be easier with autonomy, but now that you mention it, I think it will be an eye-opener for people about robotics because they've never really seen most people.

I have seen a robot and there are hundreds of thousands of Teslas, it won't be long before there are a million of them that have autonomous capability and drive themselves without a person inside and when you use it, you can see the kind of personality evolution and the thought of cars. with each iteration of

autopilot

you can see that you are not sure about this or understand it, but now it is safer, now it moves in a slightly different way, like you can tell immediately if a car is on, it tells

autopilot

because it only has small nuances of movement. it just moves in a slightly different way, it will make Aunt Ella, for example, on the highway much more accurate about being in the center of the lane than a person would if you drive down the road and look at where the human driven cars are. .

The cars are within their lane, they like bumper cars and then they move all over the place. The car and autopilot are in neutral. Yes, from the incredible work being done on your network, you are learning fast. Autonomy is still very, very difficult, we don't really know. how difficult it is, of course, you look at most problems, you approach this one with an exponential lens, but even with exponential improvement, things can take longer than expected sometimes, so where is Tesla currently in your search for total autonomy? In this sense, when can we see a successful deployment of full autonomy on the road?

The probability of an intervention is extremely low. Yes, for autonomy on the road with the latest version, especially the likelihood that intervention will be necessary. This query is actually quite low, in fact. I'd say for stop-and-go traffic, Matt was much safer than a person at this point. Let's not forget that the probability of an injury or impact is much lower for a pilot in a person and was navigating, changing lanes, taking highway junctions. and then we approach it from the other direction, which is full autonomy at low speed and in a way it's like if a person learns to drive, you learn to drive in a parking lot, you know, you know, the first time you learn to drive you probably don't out to jump onto Wolcott Street in San Francisco, that would be crazy, you're driving in the parking lot, you do things right at slow speed and then the missing piece that we're working on is the traffic lights and stuff on the streets. .

The streets of Esau are obviously relatively easy as well because you know where the street is that was geocoded and then you use the visualization to see where the line is and stop the line to light up the GPS, so in reality it's probably complex traffic lights. and windy roads are the two things to sell: what is the most difficult perception of control for these problems, so being able to perceive everything perfectly or devise a plan once you perceive everything, how to interact with all the agents of the environment in your sense. From a learning perspective, it is more difficult perception or action and then a beautiful giant multi-task learning neural network.

The most difficult thing is to have a kur representation of the physical objects in vector space, so transporting the visual input, mainly visual input, some sonar and radar, and then creating the precise vector spatial representation of the objects that surround you, once you have a vector representationaccurate, the flanker and the control is relatively easier, it's relatively easy, basically, once you have an accurate vector representation, then you are like a video game like cars. in Grand Theft Auto or something, they work pretty well, they drive down the road, they don't crash, you know, unless you crash into them, that's because they have an accurate vector-based representation of where the cars are. and they're just folded and then they show it as output, you get a high level sense that Tesla is on the way to being able to achieve full autonomy, so on the road, yes, yes, absolutely and there is still no driver status as a driver. feeling and we have the feeling of the driver talking behind the wheel, that's right, yes, by the way, just a quick comment about karaoke, most people think it's fun, but I also think it's a driving characteristic that I've been saying for a long time, singing in a car is really good for attention management and alertness uh sorry, Tesla karaoke again, it's great, it's one of the most fun features of the car.

Do you think there is a connection between fun and safety? Sometimes, yes, they are both at the same time, that's great. I just met. with and ruined wife of uh Carl Sagan oh yeah, directed cut cosmos I'm generally a big fan of Paul Sagan, he's super cool and they had a great way of bringing things, all that consciousness, all civilization, everything we've ever known and done is underway. People also have this little

blue

dot, they get too stuck there, this is like fighting between humans and this is not a big thing.image they take civilization for granted and not the continuity of existence they shouldn't do that look at history of civilizations their eyes and fall and now civilization is everything is globalized and that is why we are civilization I think that now it goes up and down together there is no there is no geographical isolation, this is a great risk, things do not always go well, like this should be, it's an important lesson from history in 1990, at the request of Carl Sagan, the Voyager 1 spacecraft, which is a spacecraft that goes further than anything created by humans in space.

I turned around to take a picture of the Earth from 3.7 billion years ago and while you're talking about the

pale

blue

dot, that image takes up less than one pixel in that image, you know, it appears as a little blue dot like a

pale

blue dot like Carl Sagan called it that he talked about this point of ours in 1994 and if you could humor me, I was wondering if in the last two minutes you could read the words that he wrote described in this buildup, sure yes, finally the universe appears to be 13.8. billion years, like the Earth, four and a half billion years, you know, another half billion years or so, the Sun will expand and probably evaporate the oceans and make life on Earth impossible, which means that If consciousness temperature had been taken, more time would be sent. to evolve, it would never have failed in all its attempts and for a longer time and I wonder how many dead civilizations from one planet there are in the cosmos that never reached the other planet and finally became extinct or were destroyed by external factors probably. some, it is only possible to try to travel to Mars barely if G were 10% higher, it wouldn't really work if empty was 10% lower, it would be easy to tear off a single stage from the surface of the module, far from the surface of the earth .

Mars is 37 cents with gravity, it's a giant blue stick, you know, channeling Costigan, look again at that dot here, which is home, it's us, everyone you love, everyone you know, everyone those you have ever heard of every human being. was once lived during their lives the aggregate of our joy and suffering thousands of safe religions ideologies and economic doctrines every hunter and gatherer every hero and coward every creator and destroyer of civilization every king and peasant every young couple in love every mother and father child hopeful inventor and explorer, every moral teacher, every corrupt politician, every superstar, every supreme leader, every saint and sinner in the history of our species, lived there on a speck of dust suspended in a ray of sunlight, our planet is a lonely spot in the great cosmic darkness that surrounds us. darkness in all this immensity there is no indication that help will come from elsewhere to save us from ourselves the earth is the only world known so far that harbors life there is no other place, at least in the near future, to which our species can migrate this is Not true, this is Mars Fault and I think Carl Sagan would agree that he couldn't even imagine it at the time, so thank you for making the world dream and thank you for speaking out today.

I really appreciate it, thank you.