text
stringlengths 28
334k
|
|---|
<s>[INST] also. He was convinced of uh yes, [/INST] he was absolutely </s><s>[INST] based on what exactly by linguistic like human based linguistic canal. She [/INST] she he whatever, you know, Kayla was ended up being like a modification of his sister's name. The real guy's sister's name was so good at building the back story all these guys and and it's funny like these guys are part of a hacking crew. They social engineer the ship out of each other just to build if one of them ever gets caught, they'll convince the everybody else that you know, they're a Brazilian uh you know, I sp owner or something like that and that's how I'm so powerful. </s><s>[INST] Well yeah, that's social engineering aspect. Is part of living a life of cybercrime or cyber security and the offensive or defensive. So a B unit can also just attention of a tangent first. [/INST] That's my favorite tangent. </s><s>[INST] Okay. Um is it possible for me to have a podcast conversation with somebody who hasn't been caught yet and because they have the conversation they still won't be caught. And is that a good idea meaning is there a safe way from criminal to talk to me on a podcast. [/INST] I would think so </s><s>[INST] I [/INST] would think that that that someone could I mean, someone who has been living a double life for, for long enough where you think they're not a criminal? Um, no, </s><s>[INST] no, no. They would have to admit that they they would say I am a V unit, [/INST] so you would want to have a conversation with the unit? </s><s>[INST] Yes. [/INST] Is </s><s>[INST] there a way I'm just speaking from an FBI perspective, technically speaking, because I I let me explain my motivation. I think I would like to be able to talk to people from all walks of life and understanding criminals. Understanding their mind, I think is very important and I think there's fundamentally something different between a criminal who is still active versus one that's been caught the mind just from observing. It changes completely once you're caught, you know, you have a big shift in your understanding of the world. Um, I mean that I do have a question about the ethics of having such conversations, but first, technically, uh is that, is it possible [/INST] if I was technically advising you? I would say, first off, don't advertise it, don't the fewer people that you're going to tell that you're having this conversation with the better. Um, and yeah, you could do it in person. Are you doing? It </s><s>[INST] would be amazing. Yeah, but their their face would not be shown. [/INST] Face would not be Yeah, I mean, you couldn't publish the show for a while, They'd have to put a lot of trust in you that you are not going to, you're going to have to alter those tapes. I say tapes because it's old school, you know. Exactly. I'm sure a lot of people just said that like ship this old guy just tape </s><s>[INST] VHS was the 1800s I think. [/INST] Um But yeah, yeah, you could do it. They have to have complete faith and trust in you that you destroy the originals after you've altered it. What </s><s>[INST] about if they don't have faith, is there a way for them to obtain security? Um so, uh like for me to go through some kind of process where I meet them somewhere where, I [/INST] mean, you're not going to do it without a bag over your head. I don't know if that's the life you want to </s><s>[INST] live. I'm fine with a bag over my head that's gonna get taken out of context. But I just, I think it's a worthy effort. It's a worthy, it's worthy to go through the hardship of that to understand the mind of somebody. I think fundamentally, conversations are a different thing than the operation of law enforcement. Understanding the mind of a criminal, I think is really important. [/INST] I don't know if you're going to have the honest conversation that you're looking for. I mean, it may sound honest, but it may not be the truth I found most times when I was talking to criminals, it's lies mixed with half truths. Uh, and you you kinda it's if they're good, they can keep that story going for long enough. If they're not, you know, you kind of see the relief in them. When you finally break that wall down, </s> |
<s>[INST] in that same way the idea that the U. S. Government or any government could be doing mass surveillance on its citizens is um is one that was a wake up call because you could imagine the ways in which that could um be like you could abuse the power of that to control the citizenry for political reasons and purposes. [/INST] Absolutely. You know, you could abuse it. I think during part of the Snowden League saw the two N. S. A. Guys were uh, modern, like their girlfriends and there's rules in place for that. Those people should be punished for abusing that. But how else are we going to hear about? You know, terrorists that are in the country talking about birthday cakes. Uh, and you know, that was a case where that, that was the trip word that, that, you know, we're gonna go bomb new york city's subway. </s><s>[INST] It's complicated, but it just feels like there should be some balance of transparency. There should be a check on that [/INST] power because </s><s>[INST] like, you know, in the name of the war on terror, you can sort of sacrifice that there is a trade opportunity, security and freedom, but it just feels like there's a giant slippery slope on the sacrificing of freedom in the name of security, [/INST] I hear you. And, </s><s>[INST] and you [/INST] know, we we live in a world where, well, I live in a world where I had to tell you exactly when I arrested someone, I had read a 50 page document of how I arrested you, uh, and all the probable cause I have against you and all that. Well, you know, bad guys are reading that they're reading how I caught you and they're changing the way they're doing things, they're changing their M. O. Um you know, they're doing it to be more secure if, you know, we tell people how we're monitoring, you know, how what we're surveilling we're going to lose that. I mean the terrorists are just going to go a different way and I'm not trying to again I'm not big government, I'm not trying to say that, you know, it's cool that that we're monitoring the U. S. Government's monitoring everything, you know, big text monitoring everything that they're just monetizing it versus uh possibly using it against you. </s><s>[INST] But there's a balance in those 50 pages. They have a lot of value um if they make your job harder but they prevent you from abusing the power of the job. There's a balance. That's a tricky balance. So the chat logs in Iceland give you evidence of the heroin and all the large scale czar level uh drug trading. What else did it give you in terms of the how to catch [/INST] gave us infrastructure. So the onion name was actually running on a server in France. So if you like, and it only community through a back channel of VPN um to connect to the Iceland server. Um there was a Bitcoin like kind of vault server that was also in Iceland. And I think that was so that the admin couldn't get into the Bitcoins the other admins that were hired to work on the site so you could get into the site but you couldn't touch the money, only ross had access to that. And then you know another another big mistake on ross's part is he had the backups for everything at a data center in philadelphia don't put your infrastructure, the United </s><s>[INST] States. [/INST] I mean again let's not make a playbook but you know </s><s>[INST] I I think the people of competence would know already. [/INST] I agree </s><s>[INST] but it's interesting that he wasn't competent enough to make so he was incompetent in certain ways. Yeah [/INST] I don't I don't think he was a mastermind of setting up and an infrastructure that would protect his his his online business because you know keeping chat logs, keeping a diary, putting infrastructure where it shouldn't be um bad decisions. </s><s>[INST] How did you figure out that he's in san Francisco? [/INST] So we had that part with Jared that he was on the west coast and then was he was a partner um in uh he was a DHS agent um worked for HSI homeland security investigations in Chicago. Uh He started his Silk Road investigation because he was working at O'hare and a weird package came in um come to find out, he traced it back to Silk Road. So he started working at a Silk Road investigation long before I started my case and he made his way up undercover all the way to be an admin on Silk Road. Um So he was talking to ross on a jabber server, the private jabber server, Private chat communication server. And we noticed that ross is um time zone on. That jabber server was set to the west coast. So we we had pacific time on there. So we had a region one 24th of the world was covered of where we thought he might be. </s><s>[INST] And from there, how do you get to san Francisco? [/INST] There was another guy an I. R. S. Agent that was part of the team and he used a powerful tool um to find his clue. Um He used the world of google. Uh he simply just went back and googled around for Silk Road at the time was coming up and found some posts on like some help forums that this guy was starting in onion website and wanted some Cryptocurrency help. And if you could help him please reach out to ross dot at gmail dot com. Um In my world, that's that's a clue. So </s><s>[INST] okay, so that that's as simple as that. [/INST] And the name he used on that post was Frosty. </s><s>[INST] Yeah, so you have to connect Frosty and other uses and Frosty and here's a gmail. And the gmail has the name the [/INST] gmail posted. That, that I need help under the name Frosty on this forum. </s> |
<s>[INST] So I I feel pretty um you know, I'm pretty short on adjectives when I describe stuff adjectives a tricky. In fact, I try to avoid adjectives because I think that's a slippery slope that you just live in the land of adjectives and you never actually make a point. Anyway. I guess you're supposed to use an audio form words to describe the thing that you're describing so that you can understand have to convert the ideas of my brain into ideas in your brain and hence the beauty of human human communication. Anyway I feel pretty good in the thing I like the way it looks like the way it feels you should check it out to right now. You can get a special discount. Just go to mizzen and main dot com and use promo code lex. This show is also brought to you by Inside Tracker a service. I used to track biological data. They have a bunch of plans. Most included blood tests that uh use machine learning algorithms on that blood data, DNA data and fitness tracker data to give you suggestions on what you should do with your [/INST] life. </s><s>[INST] It's signs back recommendations for positive diet and lifestyle changes. It's really exciting to me the kind of stuff you could do with machine learning algorithms on data that comes from the human body. But the first piece of that I think that's essential is to actually be collecting that data from you or give the algorithm enough data about you. They can make such predictions and a lot of that has to be done in a way that's uh privacy preserving and effective in terms of not just low number of samples, single instances of data collection but across the lifespan across many weeks, months years and so on. Anyway. Inside Tracker is one of the company's doing a great job of pushing that kind of idea forward. You can get special savings for a limited time and you go to inside tracker dot com slash lex This episode is also brought to you by eight sleep and its new pod three mattress. They recently launched the next generation. It's always nice when the number go up Anyway. I love the originals. I love that you want even more. [/INST] It's </s> |
<s>[INST] Okay, cool. So what, what some, what are some interesting things that came to life from this from this work? Uh, do you like, what are some insights about um, about games where natural language is involved and cooperation? Deep cooperation is involved? Well, [/INST] I think there's a few insights. Um, so first of all the fact that you can't rely purely, or even largely on self play, that you really have to have an understanding of how humans approach the game. Um, I think that that's one of the major conclusions that I'm drawing from this work. Um, and that is I think applicable more broadly to a lot of different games. So we've actually already taken the approaches that we've used in diplomacy and tried them on uh, cooperative card game called Hannah B and we've had a lot of success in that game as well. Um, on the language side, I think the fact that we were able to control the language model um, through this intense approach was very effective. Um, and it allowed us instead of just imitating how humans would communicate, were able to go beyond that and able to um, feed into its superhuman strategies that it can then um, you know, generate messages corresponding to </s><s>[INST] is there something you could say about detecting whether a person or a i is lying or not? The [/INST] bot doesn't explicitly try to calculate whether somebody is lying or not, but what it will do is try to predict what actions they're going to take, given the communications, given the messages that they sent to us. So given our conversation, what do I think you're going to do? And implicitly, there is a calculation about whether you're lying to me in that, you know, if if you're based on your messages, if I think you're going to attack me this turn. Um, even though your messages say that you're not, then, you know, essentially the body is predicting that you're lying, but it doesn't view it as as lying the same way that we would view it as lying, but </s><s>[INST] you could probably reformulate with all the same data and make a classifier lying or not. [/INST] Yeah, I think, I think you could do that um that was not something that we were focused on, but I think that it is possible that, you know, if you came up with some measurements of like what does it mean to tell a lie? Because there's there's a spectrum, right? Like if you're withholding some information, is that a lie? Um if you mostly telling the truth, but you forgot to mention this like one action out of like 10, Is that a lie? Um It's hard to draw the line, but you know, if you're willing to do that and then you could possibly use it to uh this </s><s>[INST] feels like an argument inside a relationship now. What constitutes a lie um depends what you mean by the definition of the word is okay. Um still, it's fascinating because trust and lying is all intermixed into this and it's language models that are becoming more and more sophisticated. It's just a fascinating space to explore. Um what what what what do you see as the future of this work um that is inspired by the breakthrough performance that you're getting here with diplomacy. [/INST] Uh I think there's a few different directions to take this work. Um I I think really what is showing us is the potential that language models have. I mean I think a lot of people didn't think that this kind of result was possible even today despite all the progress that we made in language models. And so it shows us how we can leverage the power of things like self play on top of language models to get um increasingly better performance and um the ceiling is really much higher than what we have right now. </s><s>[INST] Is this transferable somehow to to chatbots for the more general task of dialogue? So because there is a kind of negotiation here a dance between entities that are trying to cooperate and at the same time a little bit adversarial which I think maps somewhat to the general, you know, the entire process of reddit or like internet communication. You're cooperating your adversarial, you're having debates, you're having uh camaraderie, all that kind of stuff. [/INST] I think one of the things that's really useful about diplomacy is that we have a well defined value function. There is a well defined score that the body is trying to optimize and in a in a setting like a general chat bot setting it needs it would need that kind of um objective in order to fully leverage the techniques that we've developed. </s><s>[INST] What about like what we talked about earlier with N. P. C. S. Inside video games like how can it be used to create for elder scrolls, six more compelling um N. P. C. S. That you could talk to instead of instead of committing all kinds of violence with a sword and fighting dragons just sitting in a tavern and drink all day and talk to the chat pot. [/INST] The way that we've approached on diplomacy is you condition the language on an intent. Now that intent and diplomacy is is an action but it doesn't have to be. And you can imagine, you know, you could have N. Pcs in video games or the metaverse or whatever where there's some intent or there's some objective that they're trying to maximize and you can specify what that is. Um and and then the language can correspond to that intent. Now I'm not saying that this is you know happening imminently but um I'm saying that this is like a future application potentially of this direction of research. </s> |
<s>[INST] I don't even know how to comprehend either of those numbers frankly. Uh just a smaller side because I am a computer science person. I also find sally or thomas are beautiful. And so you have somebody like Stephen Wolfram who recently has been very excitedly exploring a proposal for a data structure. That could be um the numbers that would make you a little bit happier in terms of scale because they're very, very, very, very tiny. Um do you like this space of exploration of really thinking, letting go of space time, let go of everything and trying to think what kind of data structures could be underneath this whole mess. [/INST] That's right. So if they're thinking about these as outside of space time, then that's what we have to do, that's what our best theories are telling us. You now have to think outside of space time. Of course I should back up and say, we know that Einstein surpassed Newton, right? But that doesn't mean that they're not good work to do Newton. There's all sorts of Newtonian physics that takes us to the moon and so forth, and there's lots of good problems that we want to solve with. Newtonian physics. The same thing will be true of spacetime. Well, still, it's not like we're going to stop using spacetime will continue to do all sorts of good work there, but for for those scientists who are really looking two go deeper to actually find the next, you know, just like what Einstein did to Newton, what what are we gonna do? Einstein, how do we get beyond Einstein and quantum theory to something deeper, then we have to actually let go. And if we're going to do like this to me, to kind of approach, it's critical that it's not automata in spacetime, it's automata prior to space time, from which we're going to show how space time emerges if you're doing automated within spacetime. Well, that might be fun model, but it's not the radical new step that we need. </s><s>[INST] Yeah, so the space time emerges from that whatever system, like you're saying it's it's a dynamical system, do we even have an understanding what dynamical means when we go beyond [/INST] when </s><s>[INST] you start to think about dynamics? That could mean a lot of things. Even causality could mean a lot of things. If if we if we realize that everything is an interface uh like what how much do we really know is an interesting question because you you brought neurons I gotta ask you and another yet another danger. There's a paper remember a while ago looking at called coulda neuroscientists understand a microprocessor and I just enjoyed that thought experiment that they provided. Which is they basically it's a couple of neuroscientists eric Jonas and Conrad according who used the tools of neuroscience to analyze a microprocessor computer, computer chip, [/INST] we're leaving it here, what happens and so forth. And if you go and lesion in the computer, it's very very clear that lesion experiments on computers are not going to give you a lot of insight. </s><s>[INST] And also the measurement devices and the kind of sort of just using the basic approaches of neuroscience collecting the data, trying to intuit about the underlying function of it. And that helps you understand that our scientific exploration of concepts um depending on the field are are maybe in the very very early stages, I wouldn't say it leads us astray perhaps it does sometimes, but it's not a it's not anywhere close to some fundamental mechanism that actually makes the thing work? I don't know if you can sort of comment on that in terms of using neuroscience to understand the human mind and neurons are we really far away potentially from understanding in the way we understand The transistors. Enough to be able to build a computer. So 1 1 um one thing about understanding is you can understand for fun the other one is to understand so you could build things and and that's when you really have to understand [/INST] exactly. In fact what got me into the field that I. M. I. T. Was um work by David Marr on this very topic. So David Marr was professor at MIT but he'd done his PhD in neuroscience studying just the architecture of the brain. But he realized that his his his work it was on the cerebellum. He realized that his work as rigorous as it was left him unsatisfied because he didn't know what the cerebellum was for </s><s>[INST] and [/INST] and why it had that architecture. And so he he went to M. I. T. And he was an ai lab there and and he said he had this three level approach that really grabbed my attention. So I when I was an undergrad at U. C. L. A. I read one of his papers in a in a class and said who is this guy? Because he said you have to have a computational theory what is being computed and why? Mhm. An algorithm? How is it being computed what what are the precise algorithms and then the hardware? How does it get in stand located in the hardware? So to really do Neurosciences argued, we need to have understanding at all those levels. And that really got me I loved the neuroscience but I realized this guy was saying if you can't build it you don't understand it effectively. And so that's why I went to MIT and I had the pleasure of working with David until he died as just a year and a half later. So there was there's been that idea that, you know, with neuroscience we have to have in some sense, a top down model of what what's being computed and why that we would then go after the same thing with the trying to reverse engineer a computing system like your laptop. We really would we really need to understand what the user interface is about and why we have what are the keys on the keyboard for and so forth. You need to know why to really understand why all the circuitry and what what it's for. Now we don't evolution of natural selection does not tell us the deeper question that we're asking the answer to the deeper question which is why what what why what what's this deeper reality and what's it up to and why all it tells us is that whatever reality is, it's not what you see, What you see is just an adaptive fiction. </s> |
<s>[INST] What role do humans play in this? So what's the this biological system like a human body is made up of cells? What what role like how do you optimize the human system, the multiple engineers collaborating, figuring out what to focus on what to contribute, which which task to optimize in this neural network, who is in charge of figuring out which task needs more data. What can you can you speak to the hyper parameters the human uh system, [/INST] it really just comes down to extremely good execution from an engineering team, who knows what they're doing, they understand intuitively the philosophical insights underlying the data engine and the process by which the system improves and uh how to again like delegate the strategy of the data collection and how that works and then just making sure it's all extremely well executed and that's where most of the work is, is not even the philosophizing or the research or the ideas of it is just extremely good execution is so hard when you're dealing with data at that </s><s>[INST] scale. So your role in the data engine, executing well on it. It is difficult and extremely important. Is there a priority of like uh like a vision board of saying like we really need to get better at stoplights. Like the prioritization of tasks is that essentially, and that comes from the data [/INST] that comes to um a very large extent to what we are trying to achieve in the product for a map where we're trying to the release, we're trying to get out um in the feedback from the QA team where the system is struggling or not, the things we're trying to improve and </s><s>[INST] the QA team gives some signal some information in aggregate about the performance of the system in various conditions [/INST] and then of course all of us drive it and we can also see it, it's really nice to work with the system that you can also experience yourself and you know, it drives you home. It's </s><s>[INST] is there some insight you can draw from your individual experience that you just can't quite get from an aggregate statistical Yeah, it's so weird. Right? Yes. It's not scientific in a sense because you're just one anecdotal sample. [/INST] Yeah, I think there's a ton of uh it's a source of truth is your interaction with the system and you can see it, you can play with it, you can perturb it, you can get a sense of it, you have an intuition for it. I think numbers just like have a way of numbers and plots and graphs are much harder. Uh, it hides a lot of </s><s>[INST] it's like if you train a language model, what it's a really powerful way is, is by you interacting with [/INST] it and try </s><s>[INST] to build up an intuition. [/INST] Yeah, I think like Ellen also like he always wanted to drive the system himself. He drives a lot and uh, I want to say almost daily. So, uh, he also sees this as a source of truth. You driving the system um, and performing and </s><s>[INST] yeah, so what do you think? Tough questions here? Uh, so Tesla last year removed radar from, from the sensor suite and now just announced there's going to remove ultrasonic sensors relying solely on vision. So camera only does that make the perception problem harder or easier? I [/INST] would almost reframe the question in some way. So the thing is basically you would think that additional sensors </s><s>[INST] by the way, can I just interrupt. I wonder if a language model will ever do that If you prompted? Let me reframe your question. That would be epic. [/INST] That's the wrong </s><s>[INST] problem. Sorry. [/INST] It's like a little bit of a wrong question because basically you would think that these sensors are an asset to you. But if you fully consider the entire product in its entirety. These sensors are actually potentially liability because the sensors aren't free. They don't just appear on your car. You need suddenly you need to have an entire supply chain. You have people procuring it. There can be problems with them. They may need replacement. They are part of the manufacturing process. They can hold back the line in the production, you need to source them to maintain them. You have to have teams that write the firmware, all of the, all of it. And then you also have to incorporate and fuse them into the system in some way. And so it actually like bloats the a lot of it. And I think Ellen is really good at simplify, simplify best part is no part. And he always tries to throw away things that are not essential because he understands the entropy in organizations and an approach And I think in this case the cost is high and you're not potentially seeing it if you're just a computer vision engineer and I'm just trying to improve my network and you know, is it more useful or less useful? How how useful is it? And the thing is once you consider the full cost of a sensor, it actually is potentially a liability and you need to be really sure that it's giving you extremely useful information in this case. We looked at using it or not using it and the delta was not massive. And so it's not useful </s><s>[INST] is it also blow in the data engine like having more sense [/INST] percent </s><s>[INST] and [/INST] these sensors, you know, they can change over time. For example, you can have one type of say radar, you can have other type of radar. They change over time now you suddenly need to worry about it. Now suddenly you have a column in your sequel light telling you, oh, what sensor type was it? And they all have different distributions and then uh, they can, they just, they contribute noise and entropy into everything and they bloat stuff And also organizationally has been really fascinating to me that it can be very distracting. Um, if you, if you, if you only want to get to work is vision, all the resources are on it and you're building out a data engine and you're actually making forward progress because that is the, the sensor with the most bandwidth, the most constraints in the world and you're investing fully into that and you can make that extremely good if you're, you're only a finite amount of sort of spend of focus across different facets of the system. </s> |
<s>[INST] So that's that's an incredible book. And some of the same people you mentioned Ed Franken which I guess of expert systems, fame and Minsky of course people know in the I world but they represent the artificial intelligence [/INST] world that do hope and </s><s>[INST] dream that ai is intelligence. Is [/INST] that it was my thinking well you know I see where they're coming from and from suspected. Yeah, you're right. But that's not my perspective. So I thought I had to say it and as I was writing my book, you see, I thought well I don't really know anything about neurophysiology. What am I doing writing this book? So I certainly reading up about neurophysiology and I read and I think I'm trying to find out how it is. The nerve signals could possibly preserve quantum coherence and all I read is that the electrical signals which go along the nerves create a effects through the brain. There's no chance you can isolate it. So this is hopeless. So I come to the end of the book and I'm more or less give up. I just think of something which I didn't believe in this. Maybe this is a way around it. But no. And then you say, I thought, well, maybe this book at least stimulate young people to do science or something. And I got all these letters from old retired people instead is the only people who could have time to read my book. </s><s>[INST] So except [/INST] for Stewart Hammer off. </s><s>[INST] Except for Stuart hamm on Stuart [/INST] Hammer off wrote to me and he said, I think you're missing something you don't know about micro tubules, do you didn't put it quite like that. But that was more or less it. And he said, this is what you really need to consider. So, I thought, my God, yes, that's a much more promising </s><s>[INST] structure. So, I mean, fundamentally you were searching for the source of non computable source of consciousness within the human brain in the biology. And so what are my uh if I may ask, what are micro tubules? [/INST] Well, you see, I was ignorant and what I'd read. I never came across them in in in the books. I looked at perhaps I only read rather superficially, which is true, but I didn't know about micro tubules Stewart. I think one of the things here was impressed him about them. When you see pictures of mitosis that's a cell dividing. And you see all the chromosomes and the chromosomes get they'll get or get lined up and then they get pulled apart. And so as the cell divides the half the chromosomes go, you know how they want to divide into the two parts and they go to different ways. And what is it that's pulling them apart? Well those are these little things called micro tubules. And so he started to get interested in them. And he formed the view when he was his day job. A night job of what we call it is to put people to sleep. Except he doesn't like calling to sleep because it's different general anesthetics in a reversible way. So you want to make sure that they don't experience the pain that would otherwise be something that they feel and consciousness is turned off for a while and it can be turned back on again. So it's crucial that you can turn it off and turn it on. And what do you do when you're doing that? What do general anesthetic gasses do and see he formed the view that it's the micro tubules that they affect. And the details of why he formed that view is not. Well that's clear to me. But there but there's an interesting story he keeps talking about. But I've found this very exciting because I thought these structures these little tubes which inhabit pretty well all cells it's not just neurons apart from red blood cecil red blood cells they inhabit pretty well all the other cells in the body, but they're not all the same kind, you get different kinds of micro tubules and the ones that excited me the most. This is may still not be totally clear, but the ones that excited me most were the ones that the only ones that I knew about at the time, because they were they're very, very symmetrical structures. And I had reason to believe that these very symmetrical structures would be much better at preserving a quantum state quantum coherence, preserving the thing without you just need to preserve certain degrees of freedom without them leaking into the environment once they leak into the environment, your loss. So you got to preserve these quantum states at a level which the state reduction process comes in and that's where I think the non compute ability comes in and it's the measurement process in quantum mechanics. What's going on? </s><s>[INST] So something about the measurement process and what's going on? Something about the structure of the micro tubules. Your intuition says maybe there's something here maybe this kind of structure allows for the the mystery of the [/INST] there was a much better chance. Yes, it just struck me partly, it was the symmetry because there is a feature of symmetry, you can produce, preserve quantum coherence much better with symmetrical structures. There's a good reason for that. And that impressed me a lot. I didn't know the difference between the a lattice and be lighters at that time which could be important. Now that couldn't that's c which isn't talked about much </s> |
<s>[INST] And you were deeply involved with that [/INST] at the time. I was very much involved with that. I was like the lead architect, I ended up talking to people who had already developed serious third party libraries that did similar things and sort of taking ideas from them and getting their feedback on my design and eventually we put it in the standard library and after a few years I got distracted. I think the thing the big thing that distracted me was actually type annotations, </s><s>[INST] but [/INST] other people kept it alive and kicking and it's been quite successful actually in the world of python web clients. </s><s>[INST] So initially what are some of the design challenges there in that debate for the pep And what are some things that got rejected? What are some things that got accepted to stand out to you? [/INST] There are a couple of different ways you can handle parallel i O. And this happens sort of at an architectural level in operating systems as well, like Windows prefers to do it one way and UNIX prefers to do it the other way. You sort of you have an object that represents a network endpoint, say connection with a web browser that your client and say you're you're you're waiting for an incoming request. Two fundamental approaches are okay, I'm waiting for an incoming request. I'm doing something else. Come wake me up or of course sort of come tell me when something interesting happened like a packet came in on that network connection. And the other paradigm is we're on a team of a whole bunch of people with maybe a little mind and we can only manage one web connection at a time. So I'm just sitting looking at this this web connection and I'm just blocked until something comes in and then uh I'm already waiting for it, I get I get the data, I process the data and then I go back to the top and say now sort of I'm waiting for the next packet. Those are about the two paradigms. One is a paradigm where there is sort of notionally a threat of control, whether it's an actual operating system, thread or more an abstraction in, we call them tasks, but the task in a or a threat in other contexts is devoted to one thing and it has logic for all the stages, like when it's a web request, like first wait, wait for the first line of the web request parse it because then, you know, if it's a get or a post or put or whatever or an error, uh then wait until you have a bunch of lines until there's a blank line and parts that as heathers and then interpret that and then wait for the rest of the data to come in. If there is any more that you request, expect that sort of standard web stuff. And the other thing is, and there's always endless debate about which approach is more efficient and which approach is more error prone where I just have a whole bunch of stacks in front of me and whenever a packet comes in, I sort of look at the number of the fact that there's some number on the packet and I say, oh that packet goes in this pile and then I can do a little bit and then sort of that pile provides my context. And as soon as I'm done with the processing, I sort of I can forget everything about what's going on because the next packet will come in from some random other client and it's that pile, </s><s>[INST] this pile [/INST] uh and every time a pile is maybe empty or full or whatever the criteria is, I can toss it away or use it for a new space but several traditional third party libraries for asynchronous I o processing in python chose the model of a callback and that's that's the idea where you have a bunch of different stacks of paper in front of you and every time someone gives you a piece gives you a new sheet, you decide which stack it belongs to. And that leads to a certain style of spaghetti code that I find sort of aesthetically not pleasing. And I was sort of never very successful. And I had heard many stories about people who were also sort of complaining about that style of coding. It was very prevalent in javascript at the time, at least because it was like how the javascript event loop basically works. And so I thought well the task based model where each task has a bunch of logic. We had mechanisms in the python language that we could easily reuse for for that. And I thought I want to build a whole library for asynchronous networking i o uh and all the other things that may need to be done a synchronously uh based on that paradigm and so I just chose a paradigm and try to see how far I could get with that. And it turns out that it's pretty good paradigm. So </s><s>[INST] people enjoy that kind of paradigm programming for asynchronous io, relative to callbacks. Okay, beautiful. So how does that all interplay with the infamous guild, the go, the global interpreter lock, maybe can you say what the deal is and how does it dance beautifully with a sen ko [/INST] The global interpreter lock solves the problem that python originally was not written with either asynchronous or or parallelism in mind at all. There was no concurrency in the language, there was no parallelism, there were no threads, only a small number of years into python's initial development, all the new cool operating systems like uh Sun. Os and silicon graphics, I rex and then eventually pas IX and Windows all came with threading libraries that lets you do multiple things in parallel and there is a certain certain sort of principle which is the operating system handles the threads for you and the program can pretend that there are as many cpus as as there are threats in the program uh and those Cpus were completely independently and if you don't have enough Cpus operating system sort of simulates those extra Cpus on the other hand, if you have enough Cpus, you can get a lot of work done by deploying those multiple Cpus, but python wasn't written to to do that. Uh And so as libraries for for multi threading were added to C. But every operating system then there was adding their own version of that. We thought and maybe we were wrong, but at the time we thought, well we quickly want to be able to support these multiple threads Because they seemed at the time in the early 90s when they were new, at least to me, they seemed a cool, interesting programming paradigm. And one of the things that that python at least at the time felt was nice about the language was that we could give a safe version of all kinds of cool new operating system toys to the python programmer. Like, I remember one or two years before threading, I I had spent some time adding networking sockets uh to python and there were very literal translation of the networking sockets that were in the BSD operating system. So UNIX BSD But the nice thing was if you're using sockets from python, then all the things you can do wrong with sockets in C would automatically give you a clear error message instead of just ending up with a malfunctioning hanging program. And so we thought, well we'll do the same thing with threading, but we didn't really want to rewrite the interpreter to be threat safe because that that was was like there would be a very complex re factoring of all the interpreter code and all the runtime code because all the objects were written with the assumption that there's only one thread. And so we said, okay, well we'll take our losses will provide something that looks like threads and as long as you only have a single CPU on your computer, which most computers at the time did uh it feels just like threats because The whole idea of of multiple threats in the Os was that even if your your computer only had one CPU you could still fire up at many threads as you wanted. Well, within reason maybe 10 or 12, not 5000. Uh And so we thought we had conquered the the abstraction of threats pretty well because multi core Cpus were not in, in most python programmers hands anyway. And then of course a couple of more iterations of moore's law and computers getting faster. And at some point, uh the chip designers decided that they couldn't make the Cpus faster, but they could still make them smaller and so they could put multiple Cpus on one chip. And suddenly there was all this pressure about do things in parallel and that's where the the solution we had in python didn't work. And that's that's sort of the moment that the gill became became infamous because the guilt, the guilt was the solution, we used to sort of take this single interpreter and share it between all the different operating system threats that you could create. And so as long as the hardware physically only had one CPU that was all fine. And then as hardware offenders were suddenly telling us all, oh, you got to paralyze everything's got to be paralyzed. People started saying, oh but we can use multiple threads in python and then they discovered, oh, but actually all threats run on a single single core. </s> |
<s>[INST] Yeah, I was just, uh, as I often do, rereading Viktor Frankl's man's search for meaning. And he said something, Um, I just I often pull out sort of lines. The mere knowledge that a man was either a camp guard or prisoner tells us almost nothing. Human kindness can be found in all groups, even those which as a whole it would be easy to condemn. So that's speaking to you Feel for those people at the lowest level, implementing the orders, Um, [/INST] of those above right. And also you were yourself What will happen if you were given those same orders, You know, I mean, what would you do? What kind of reaction would you have in a similar situation? And, you know, you don't know </s><s>[INST] I could see myself in in World War Two, fighting for almost any country that I was born in. There's a love of community. There's a love of country that's just, uh, at least to me, it comes naturally. Just love of community and country is one such community, and I could see fighting for that country especially when you're sold the story that you're fighting evil. And I'm sure every single country was told that story effectively. And then, uh, when you're in the military and you have a gun in your hand or you're in the police force and you're ordered, go to this place and commit violence, it's hard to know what you would do. It's a mix of fear. It's a mix of. Maybe you convince yourself, you know, what can one person really do? And over time it's again that slippery slope because you could see all the people who protest who revolt, they're ineffective. So, like if you actually want to practically help somehow you're going to convince yourself that you can't one person can't possibly help, and then you have a family. So you want to make you know you want to protect your family. You tell all these stories, and over time it you naturally convince yourself to dehumanize the other. Yeah, I think about this a lot, Um, mostly because I worry that I would be a good German. [/INST] Yeah, No, no, that's right. That's right. And one of the, you know, one of my tasks as a as a teacher, right? Our students and I have classes on genocide, and I have one now, Uh, and another one, by the way, on Stalin, Um, but the one on on genocide, you know, one of my task is to try to try to get to get the students to understand This is not about weird people who live far away and in time and in place. But it's about them, you know, and that, you know, that's a hard lesson, but it's an important one. You know that that this is in all of us. You know, it's in all of us, and there's nothing, you know, and you just try to gird yourself up, you know, to try to figure out ways that maybe you won't be complicit, um, and that you learn how to stand by your principles. But it's very hard. It's extremely difficult, and you can't. The other interesting thing about it is not predictable. They've done a lot of studies of polls. For example, who during the war saved Jews, you know well, who are the Poles who saved Jews versus those who turned them in? It's completely unpredictable, you know. Sometimes it's the worst anti Semites who protect them because they don't believe they should be killed. Right? Um, and sometimes, you know, it's not predictable. It's not as if the humanists among us, you know, are the ones who, uh, you know, can consistently show up. You know, uh, you know and experience danger, in other words, and are ready to take on danger to defend your fellow human beings? Not necessarily. I mean, sometimes simple people do it, and sometimes they do it for really simple reasons. And sometimes people you would expect to do it, don't you know? And you've got that mix and it's just not predictable. </s><s>[INST] One thing I've learned in this age of social media is it feels like the people with integrity and the ones who would do the right thing are the quiet ones. Uh, in terms of humanists, in terms of activists, there's so many points to be gained, um, of declaring that you would do the right thing. It's a simple, quiet folks, because I've I've seen quite on a small, obviously much smaller scale, just shows of integrity and character when they were sacrificed to be made and it was done quietly, Uh, this sort of the small heroes, Uh, those are your right there. It's surprising, but they're often quiet. That's why I'm distrustful of people who kind of proclaim that they would do the right thing [/INST] right. And there's different kinds of integrity to I mean, I edited a memoir of a polish, uh, you know, underground fighter member of the underground who was in Majdanek in the concentration camp at Majdanek. You know, when he was just an interesting mix of different kinds of integrity. You know, on the one hand, you know, he really bothered him deeply when Jews were killed, were sent to camp, or is that sort of thing? On the other hand, he was something of an anti Semite, you know? And he would, uh you know, sometimes if Jews were his friends, he would help them. And if they weren't sometimes he was really mean to them, you know? And you could in their various levels, You know, a concentration camp is, uh, a terrible social experiment in some ways. Right? Uh, but you learn a lot from how people behave. And what you see is that you know, people behave sometimes extraordinarily well in some situations and extraordinary poorly in others, and that it's mixed and you can't predict it. And it's hard to hard to find consistency. I mean, that's the other thing. It's, you know, No, I think we claim too much consistency for the people we study and the people we think about in the past, you know, they're not consistent anymore than you. We are consistent, right? </s> |
<s>[INST] Fascinating to think that COVID-19 is uh as a result of a virus evolving together with like twitter, like figuring out how we can sneak past the defenses of the humans. So it's not bad enough. And then then the misinformation, all that kind of stuff together is operating in such a way that the virus can spread effectively. I wonder, I mean obviously uh virus is not intelligent, but there's a uh there's a rhyme and rhythm to the way this whole evolutionary process works and creates these fascinating things that spread throughout the entire civilization. [/INST] Absolutely. It's um yeah, I'm I'm completely fascinated by this idea of social media in particular how it replicates how it grows. I've been how it how it actually starts interacting with the biology of the virus masks who's going to get vaccinated politics. Like these seem so external to virus biology, but it's become so intertwined and and it's interesting and I actually think we could find out that the virus actually becomes obviously not intentionally, but you know, we could find that choosing people choosing not to our masks, choosing not to counter this virus in a regimented and sort of organized way effectively gives the virus more opportunity to escape. We can look at vaccines, you know, we're about to, we're about to have one of the most aggressive vaccination programs the world has ever seen. But we are unfortunately doing it right at the peak of viral transmission when millions and millions of people are still getting infected. And when we do that, that just gives this virus so many more opportunities. I mean, orders of magnitude more opportunity to mutate around our immune system. If we were to vaccinate everyone when there's not a lot of virus, then there's just not a lot of virus. And so there's not going to be as many, you know, I don't even know how many zeros are at the end of however many viral particles there are in the world right now, you know more than quadrillions, you know, and so if you assume that at any given time somebody might have trillions of virus in them and any given individual, so then you know multiply trillions by millions and you know, you get a lot of viruses out there and if you start applying pressure, ecological pressure to this virus that that when it's not abundant, I had the opportunity for a virus to sneak around immunity, especially when all the vaccines are identical essentially. Um it </s><s>[INST] sticks is one to mutate and then jumps takes [/INST] one in the whole world, you know, and we have to, we have to not forget that this particular virus was one, it was one opportunity and it has spread across the globe and there's no reason that can't happen tomorrow. I knew, you know it's scary. </s><s>[INST] I have a million other questions in this direction, but I'd love to talk about one of the most exciting aspects of your work which is testing or rapid testing. You wrote a great article in time On November 17. This is like a month ago about rapid testing titled How we can stop the spread of COVID-19 by Christmas, let's jot down the fact that this is a month ago. So maybe your timeline would be different, but let's say in a month. So you've talked about this powerful idea for quite awhile. Throughout the COVID-19 pandemic. How do we stop the spread of COVID-19 in a month. [/INST] Well, we uh we use tests like these, you know, So so the only reason the virus continues spreading is because people spread it to each other. This isn't this isn't magic. Um and so there's a few ways to stop the virus from spreading to each other. And that is uh you either can vaccinate everyone and vaccinating everyone is a way to immunologically prevent the virus from growing inside of somebody and therefore spreading. We don't know yet actually if this vaccine, if any of these vaccines are going to prevent onward transmission. So uh so that may or may not serve to be one opportunity. Certainly I think it will decrease transmission. But the other idea that we have at our disposal now, we had it in May, we had it in june july august september october november. And now it's december. We still have it. We still choose not to use it in this country and in much of the world. And thats rapid testing that is giving its empowering people to know that they are infected and giving them the opportunity to not spread it to their loved ones and their friends and neighbors and whoever else. Um We could have done this. We still can today. We could start we have millions of these tests. These tests are simple paper strip tests. They are uh inside of this thing is just a little piece of paper. Um No and I can actually open it up here. There we go. So this this is how we do it right here. We have this little paper strip test. This is enough to let you know if you're infectious With somewhere around the order of 99% sensitivity, 99% specificity. You can know if you have infectious virus in you, If we can get these out to everyone's homes, build these, make 10,000,000, million, 30 million of them a day. We make more bottles of Dasani water every day. We can make these little paper strip tests and if we do that and we get these into people's homes so that they can use them twice a week. Then we can know if we're infectious, you know, is it perfect? Absolutely not. But is it near perfect. Absolutely. You know, and so if we can say, hey, the transmission of this is, you know, for every 100 people that get infected right now, They go on to in fact maybe 130 additional people. And that's exponential growth. 100 becomes 130, a couple days later that 100 30 becomes Another another 100 65 people have now been infected. And you know go over three weeks and 100 people become 500 people infected. Now it doesn't take much to have those 100 people, not in fact 130, but in fact 90 all we have to do is remove say 30 40% of new infections from continuing their spread. And then instead of exponential growth you have exponential decay. So this doesn't need to be perfect. We don't have to go from 100 to 0. We just have to go and have those 100 people infect 90 and those 90 people. In fact, you know, 82 whatever it might be and you do that for a few weeks and boom, you have now gone instead of 100 to 500 you've gone from 100 to 20. It's not very hard. And so the way to do that is to let people know that they're infectious. I mean I were perfect example right now I I this morning I use these tests To make sure that I wasn't infectious. Is it perfect. No, but it reduced my odds 99%. I already was at extremely low odds because I've spent my life quarantining these days. </s> |
<s>[INST] it's complicated. I mean actually might is interesting because it's I would say M. I. T. Is different than stanford for example. I think it has a more kind of [/INST] narrow um about an algae in my apartment. Well </s><s>[INST] no, it's interesting because I would argue that there's some aspect of a brand like taliban, Emmett, no relation that has a kind of interact like the brand results in the behavior of the like enforces a kind of behavior and the people and the people feed the brand. And like there's a lot, I think stanford is a good example of something that's more distributed. There's sufficient amount of diversity in like all kinds of like centers and all that kind of stuff that the the brand doesn't become one thing mitt. So engineering, it's, you know, I think that [/INST] the taliban scratching scratching for it because I think it was more like mitt than than you might imagine. But, but isn't taliban, isn't it? Pretty. I don't </s><s>[INST] think there's a diversity. [/INST] So yeah, sorry, so just rephrase so, so people say, oh the Deobandi school and like what is that? I mean, but the taliban are, they're an ethnic movement. They represent a vision of postion power. Right? Punches are people who are quite internally diverse who actually speak multiple dialects of pashto who reside across the frontier of Pakistan in Afghanistan, there are pashtuns who live all over the planet, right? There's a community Moscow California everywhere. Right? So it's a global diaspora of sorts, passengers have a kind of geological imagination. So that lots of passions can tell you the names of their grandparents, great grandparents and so on and that's kind of a, there's a sense of pride in that posture. Language is a kind of a core element of that identity, but it's not universal. So for example, you can meet people who say I am pashtun, but I don't know pashto. So as you, as you claw away at this idea, it's amorphous. It also means different things, different people at different times. So saying the taliban are postion requires lots of qualifiers because lots of passions will say no, no, I have nothing to the taliban, I hate those people. So the taliban tried to mobilize other past tunes with limited success, but their core membership is almost exclusively pashtun and they say no, no, we represent afghans, we represent pious Muslims and so in recent 23 years they've gone further to say no, we have ethnic groups, we have those bags, we have Tajiks, we have Hazaras and in the north of Afghanistan in recent years they did do a bit better drawing in people who were very disaffected because the government and they were able to diversify their ranks somewhat. But if you watched almost 15 and who they've appointed what language they've used, how if present themselves, it's clear that you know, they are pashtun, they are male and they are extremely ideologically cohesive and disciplined, I'd say, right, so I think that a lot of the polish centrism, la block some of that stuff was the way to fight a war. Um they are, they are finally, you know, a guerilla movement, they see themselves as kind of pius Robin Hoods, the rhetoric is very much about taking from the rich, taking the privilege giving to the poor being on the side of the underdog, fighting against evil. And so I mean they're bag, if you like their thing, their, their central theme, their brand is about public morality and so their origin story going back 24 is that they interceded, they broke up a gang of criminals who were trying to write people and so there's a wrenching kind of like, it's not like sexuality and on, on public morality really been the core of like, you know, we're gonna restore order and morality and how that translates into governance is something they've never sorted out. I mean, how do you run a banking system if you're internal priorities are really about, you know, the length of beard and then, and then their path to power and in a kind of abstract sense, I mean a lot of that was very much driven by, um, if you like propagating the problems of martyrdom. And that sounds, I don't mean to say that in a way that to make it sound ridiculous, make it sound like it's um, you know, a moral judgment is simply, I think the fact it's a fact of their appeal that they promised young men who have known nothing else but studying in certain schools if at all, but they've known fighting and they've known they've known victimization and this isn't I'm not asking for like sympathy for them. I think the reality is that a lot of the we know about the kind of foot soldiers is that they they lost families and bombings um in airstrikes in night raids. Yeah I mean orphans have always been a stream um living in in all male society not knowing girls not knowing women. Hearing things from outside about places like Kabul. And so there's always been this kind of urban rural dimension. It's not it's not just that but I think there's a there's a whole imagination that being taliban captures and the whole marginal thing is really it's um yeah I think to any religious person, I mean it's not a it's not a bizarre idea. I mean it animates I mean so many global traditions. You know I think the but you try to tell like an army colonel try to here to have a conversation with you know U. S. Marine about this. I mean someone get it from their own religious backgrounds but I think the it it's an alien idea but I think it is essential to kind of stretch of imagination understand that's that's attractive and now one of the dilemmas going forward is that they've got a pivot from martyrdom and some have been some have told foreign journalists. I mean it's good that we're in charge now we're going to build a proper state, but i i it's kind of boring. Um I want to keep fighting, I want to maybe I'll do it in Pakistan. </s> |
<s>[INST] so you're thinking including something bigger than just liberal education, also including science engineering and technology. That should also comment that, you know, I mostly stay out of politics and out of some of these aspects of liberal education, that's kind of been the most controversial and difficult within the university. But there is a kind of ripple effect of fear within that space into science and engineering and technology that I think has uh, has a nature that's difficult to describe. It doesn't have a controversial nature. It just has a nature of fear where you're not, You know, you're not just saying stupid stuff as a young 20 year old, you know, for example, Deep learning, machine learning is really popular in the computer science now as an approach for creating artificial intelligence systems. It's it is controversial in that space to say that anything against machine learning, saying sort of, exploring ideas that saying this is going to lead to a dead end. Um, now that that takes some guts to do as a young 20 year old within uh, within the classroom, to think like that, to raise that question in a machine learning course. It sounds ridiculous because it's like who's going to uh complain about this, but the fear that starts in uh in a course on history or on some course that cover society. The fear ripples and affects those students that are asking big out of the box questions about engineering, about computer science. And there's a lot, you know, there's like linear algebra that's not going to change, but then there's like applied linear algebra, which is machine learning and that's when robots and real system touch human beings. And that's when you have to ask yourself these difficult questions about about humanity, [/INST] even in the engineering and science and technology courses, and these are not separate worlds in two senses, I've just taken delivery of my copy of the book that eric Schmidt and Henry Kissinger have co authored on artificial intelligence. The central question of which is what does this mean for us broadly, but they're not separate worlds, you know, in C. P. Snow's sense of, you know, the chasm between science and arts because on a university campus, everything is contagious from a novel coronavirus to the behaviors that are occurring in the english department. Those behaviors if denunciation becomes a norm, you know, undergraduate denounces professor teaching assistant denounces undergraduate. Those behaviors are contagious and will spread inexorably first to social science and then to the natural sciences. And I think that's that's part of the reason why when this started to happen when we started to get the origins of disinfect dis invitation and cancel culture. It was not just a few conservative professors in the humanities who had to worry. Everybody had to worry because eventually it was going to come even to the most apparently hard stem part of the campus. It's, it's contagious. This is something Nicholas christakis should look at because he's very good at looking at the way in which social networks, like the ones that exist in a university can spread everything. But I think when we look back and ask why did work is um, spread so rapidly and rapidly out of humanity's into other parts of universities. And why did it spread across the country and beyond the United States? To the other english speaking universities, it's because it's a contagion. Uh, and these behaviors are contagious. The president of a university, I won't name said to me that he receives every day at least one denunciation, one call for somebody or rather to be fired for something that they said, that's the crazy kind of totalitarian ism light that now exists in our universities. And of course the people who want to downplay this say, oh well they're only have been 100 and something in dis invitations or oh, there really aren't that many cases. But the point is that the famous events, the events that get the attention are responsible for a general chilling. That as you say, spreads to every part of the university and creates a a very familiar culture in which people are afraid to say what they think. Self censorship. Look at the heterodox Academy data on this grows and grows. So now a majority of students will say, this is clear from the latest heterodox academy surveys, we are scared to say what we think in case we get denounced, in case we get canceled. Well, that's just not the correct atmosphere for a university in a free society. To me, what's really creepy is how many of the behaviors I see on university campuses today are reminiscent of the way that people used to behave in the soviet union or in the soviet bloc or in mao's china, the sort of totalitarian ism light that I think we're contending with here, which manifests itself as denunciations. People informing on superiors, some people using it for career advantage, other people reduced to helpless, desperate apology to try to exonerate themselves. People disappearing metaphorically, if not literally all of this is so reminiscent of the totalitarian regimes that I studied earlier in my career, that it makes me feel sick. And what makes me really feel sick is that the people doing this stuff, the people who write the letters of denunciation are apparently unaware that they're behaving exactly like people in Stalin's soviet union, they don't know that. So they clearly have there's been a massive educational failure if somebody can write an anonymous or non anonymous letter of denunciation and not feel shame. I mean, you should feel morally completely contaminated as you're doing that. But people haven't been taught the realities of totalitarian ism. For all these reasons, I think you need to try at least to create a new institution where those pathologies will be structurally excluded. </s> |
<s>[INST] molecule map of what happens to skin [/INST] which has </s><s>[INST] never been done before. So, there what are [/INST] interesting surprises there. </s><s>[INST] So [/INST] one of the things we can </s><s>[INST] see the part of what's [/INST] driving information, we can actually see </s><s>[INST] macrophages and [/INST] other dendritic cells, pieces like cells that are part of the immune system kind of creeping along towards the surface of the </s><s>[INST] skin, which is now we know it's actually [/INST] physically driving the immune system is these cells </s><s>[INST] going and creating this inflammation, [/INST] which is what leads to some of </s><s>[INST] the rashes. But we didn't [/INST] see as much in them as we saw. For example, some of the signatures of scott kelly. So we can see within the crew </s><s>[INST] who's getting more [/INST] of a rash or not or who didn't experience </s><s>[INST] any rash [/INST] and some people had changes in </s><s>[INST] vision. Some people had, [/INST] you know, other G. I. Problems even, you know, even looking at what happens to the gut </s><s>[INST] and looking at the microbiome in [/INST] the gut, other people didn't. So we were able </s><s>[INST] to see it start [/INST] to get a little bit predictive about their </s><s>[INST] medicine right now we're just diagnosing. [/INST] But it'd be good to say uh if you're going into space, </s><s>[INST] here's exactly what you [/INST] need for each bacteria </s><s>[INST] in your body. [/INST] Here's what you can really take to get rid of nausea or other ways </s><s>[INST] we can monitor to keep keep the [/INST] information down. What does it take to prepare for one of these missions? Because you mentioned, some of the </s><s>[INST] folks are not [/INST] necessarily lifelong astronauts, </s><s>[INST] you're talking about [/INST] more and more regular civilians. </s><s>[INST] What's [/INST] what does it take physiologically and psychologically </s><s>[INST] to prepare for these. They have to do a [/INST] lot of the same training that most astronauts do. So a lot of it's on hawthorne at SpaceX headquarters, which if you can </s><s>[INST] never get a chance to do a tour as [/INST] fabulous as you can see. Always john rockets being built and then we're drawing blood over there </s><s>[INST] right next to them. So it's a really [/INST] cool place. But </s><s>[INST] the, the training, they have [/INST] to go through a lot of the ops, a lot of the programming just in </s><s>[INST] case most of the systems are [/INST] automated on the Dragon </s><s>[INST] and other spacecraft, [/INST] but just in case, so I have to go through all </s><s>[INST] the majority of training [/INST] if you want to go to the </s><s>[INST] space station, as [/INST] the actual emissions are including john Schaffner, you have to do training for some of the Russian </s><s>[INST] modules. And if you don't do that training, then [/INST] you're not allowed to go to the Russian </s><s>[INST] part of the space station apparently. [/INST] So right now john schaffner for example, unless he completes this additional training all in Russian, </s><s>[INST] he's not allowed [/INST] to learn enough Russian to be, you know, just so </s><s>[INST] it's not just technical [/INST] enough enough </s><s>[INST] enough enough. [/INST] And uh, so it's a, and </s><s>[INST] if he doesn't learn, he can't [/INST] go to that part of the space station. So interesting things like that. But you'll be, you know, it's not that far like, oh, I can see </s><s>[INST] it right there, I can't float [/INST] over to that, that capsule, </s><s>[INST] but technically [/INST] he can't go. So, you know, </s><s>[INST] is there um, [/INST] chinese </s><s>[INST] components to this, the international space [/INST] station is their collaboration there, </s><s>[INST] sadly. No, they're building [/INST] their own space station. I'm glad they're building a space station actually, eventually there'll </s><s>[INST] Be probably four space [/INST] Stations in orbit by 20, some from the orbital reefs, some from Lockheed Martin </s><s>[INST] of course axiom is, [/INST] is far ahead </s><s>[INST] right now. They're probably gonna be done first. But the [/INST] extraordinary thing </s><s>[INST] is [/INST] uh, there's </s><s>[INST] unfortunately there's no [/INST] collaboration between the, you see that as a negative. That's not the positive </s><s>[INST] kind of competition. It's it's a good [/INST] question. So </s><s>[INST] maybe [/INST] for example when we get different Nasa grants, you apply for a </s><s>[INST] grant, you get to the [/INST] lab, it goes, you know, to </s><s>[INST] go through Cornell the grants [/INST] office. I have to sign as a scientist as the </s><s>[INST] P. I. On the on the mission. Say I promise [/INST] I will move no funds or resources or any </s><s>[INST] staff to anyone in china or work with anyone in china with [/INST] these dollars that you're giving to the </s><s>[INST] lab for this mission. [/INST] And so like every </s><s>[INST] other grand I get from the [/INST] Nasa D. O. D. Or </s><s>[INST] sorry, do you let me go back to [/INST] that every other grant </s><s>[INST] I get from say the NIH or [/INST] the NSF </s><s>[INST] even symptoms D. O. D. You don't have to promise that you won't talk to anybody in china [/INST] about it. But for Nasa alone is </s><s>[INST] congressional mandated [/INST] you have to promise and sign all the paperwork that I can't do </s><s>[INST] anything with anyone in china about this. [/INST] And what if </s><s>[INST] you have said about that is I want [/INST] to at least be able to chat </s><s>[INST] with him about it and know what [/INST] they're up to, but we can't, you know, </s><s>[INST] even go to a conference in China technically with [/INST] Nasa funds about, say, space medicine or </s><s>[INST] engineering a new rocket. [/INST] I can't </s><s>[INST] go, I could go with personal [/INST] funds, but I can't use those funds. </s><s>[INST] Like you should be able to go to a [/INST] conference and in a </s><s>[INST] friendly way, talk shit [/INST] scientists, </s><s>[INST] like we're doing like, like the way scientists do really well, which is [/INST] like a compliment, </s><s>[INST] but it's a backhanded compliment. Like, like [/INST] you're doing a really good </s><s>[INST] job here and [/INST] then you kind of implied that you're doing </s><s>[INST] much better job. That's [/INST] the core of </s><s>[INST] competition. You get jealous and [/INST] then everybody's trying to improve. But that </s><s>[INST] you're ultimately talking, [/INST] you're ultimately collaborating closely. You're competing </s><s>[INST] closely as opposed to in your own silos. Well, let me ask, [/INST] uh, </s><s>[INST] in terms of [/INST] preparing um, for spaceflight, you know, I I tweeted about this and I </s><s>[INST] I joked about it and [/INST] I I talked to you along </s><s>[INST] quite a lot these [/INST] days. What I tweeted </s> |
<s>[INST] You know, To me, it's fascinating for being humble. Like the humans cannot possibly be the most amazing thing that such if you look at the entirety of the system that leads small and you paint that cannot possibly be the most amazing thing that process generates. So like if you look at the evolution, what's the equivalent in the cosmological evolution and its selection for technology, the equivalent of the human eye or the human brain universes that are able to do some, like they don't need the damn stars. They, they're able to just do some incredible generation of complexity fast on like much more than if you [/INST] think about it's like most of our universe </s><s>[INST] is pretty freaking boring. There's not much going on. It's a few rocks flying around and there's some like apes that are just like um doing podcasts, something weird on some weird planet. It just seems very inefficient. If you think about like the amazing thing, the human eye, the visual cortex can do the, the brain nervous of everything that makes us more powerful than single cell organisms. Like if there's an equivalent of that for universes there, like the richness of physics that could be uh that could be expressed through a particular set of parameters. Like, I mean that like for me, I'm so from a computer science perspective, huge fan of cellular automata, which is a nice sort of pretty visual way to illustrate how different laws can result in uh drastically different levels of complexity. So like it's like, yeah, okay, so we're all like celebrating, [/INST] look our little cellular tom and is able to generate </s><s>[INST] pretty triangles and squares and therefore we achieve general intelligence. And then there will be like some badass chuck Norris type, like uh universal turing machine type of cellular automata, they're able to generate other cellular automata that does any arbitrary level of computation off the bat it like those have to then exist. And then we're just like this, we're just, we'll be forgotten. Is this story, this is uh, this podcast just entertains a few other apes for for [/INST] a few months. Well, I'm kind of surprised to hear your cynicism. I'm very, I usually think of you as like a one who celebrates humanity in all its forms and things like that and I guess I just, I don't, I see it the way you just described. I mean, Okay, five, we've been here for 13.7 billion years and you're saying, gosh, that's a long time. Let's get on with the show already. Some other universe could have kicked our butt by now, but That's putting a characteristic time. I mean, why is 13.7 billion a long time? I mean, compared to compared to what I guess. So when I look at our universe, I see this extraordinary hierarchy that has developed over that time. So at the beginning it was a chaotic mess of you know, some plasma, nothing interesting going on there. And even for the first stars to form that a lot of really interesting uh evolutionary processes had to occur by evolutionary in that sense. I just mean, um taking place over extended periods of time and structures are forming then. And then it took that first generation of stars in order to produced the medals that uh, that then can more efficiently produce another generation of stars. We're only the third generation of stars. So we might still be pretty quick to the to the game here. So, but I don't think I don't Okay, so then, so then you have these stars that you have solar systems on those solar systems, you have, um, rocky worlds, you have gas giants, like all this complexity and then you start getting life and the, the complexity that's evolved through the evolutionary process in life forms is just, it's not a letdown to me just </s><s>[INST] and there's some of it is like some, some of the planet's is like, I see like different flavors of ice cream there. I see, but there might be water under all kinds of life forms and some volcanoes, all kinds of weird stuff. No, no, I, I don't, I think it's beautiful. I think our life is beautiful and I think it was designed that by, by design the scarcity of the whole thing. I think mortality as terrifying as it is. As fundamental to the whole reason we enjoy everything. No, I think it's beautiful. I just think that all of us, um conscious beings in the grand scheme of basically every scale will be completely forgotten. [/INST] Well that's true. I think everything is transient and that would go back to maybe something more like loud to the Tao Beijing or something where it's like, yes, there is nothing but change, there is nothing but emergence and dissolve and that that's it. But I just in this picture of this hierarchy that's developed, I don't mean to say that now it gets to us and that's the pinnacle. In fact, I think the, at a high level the story I'm trying to tease out in my research is about, Okay, well, so then what's the next level of hierarchy? And if and if it's okay, we're kind of pretty smart. I mean, talking about people like lee small and alan goose max, Take Mark. Okay, we're really smart talking about me. Okay, we're kind of, we can find our way to the grocery store or whatever, but but what's next? You know, I mean, what if, what if there's another level of hierarchy that grows on top of us that is even more profoundly capable? And I mean, we've talked a lot about super conducting sensors. Imagine these uh cognitive systems far more capable than us residing somewhere else in the solar system off of the surface of the earth, where it's much darker, much colder, much more naturally suited to them. And they have these sensors that can detect single photons of light uh, from radio waves out too, all across the spectrum to gamma rays and just see the whole universe. And they just live in space with these massive um collection optics so that they what do they do? They just look out and and experience that that vast array of of what's being developed. </s> |
<s>[INST] Yeah, I don't find it super interesting. Well, he also wrote the book, [/INST] The Real Anthony Fauci, is that the name of the book? Did he write that? I don't know, I'm not sure about that. </s><s>[INST] I'm aware of that book, I didn't know he wrote it. I think I need it, but it's been on my reading list to get, I've been trying to get a good, balanced reading list about the COVID pandemic to understand what the hell happened. And anytime I start to try to go into that place, [/INST] it's just, I'm exhausted, exhausted by it. Well, it's interesting to me that you wouldn't wait longer before delving into those books </s><s>[INST] to have maybe a more clear hindsight. But I think this is a pretty good time. You don't think so? Like this is, it depends on your goals. If you're thinking of it as a historical event, yes, you should probably wait longer. But if you're thinking about like understanding what is broken about our system that we responded so poorly, that there was so much division, what is broken about our political system that it didn't unite us, it divided us. Who's to blame? There's probably a lot of different narratives, but I feel like the more you learn about this, the better you can understand. I read, on just Vladimir Putin, I read like five biographies already, maybe more. Just, it helps to really understand the people involved, the organizations involved, I don't know, everything from the scientists to the political leaders. It felt like the blog posts and the tweets didn't quite capture the story. [/INST] They didn't quite, they didn't quite, no, one of the things I read a ton, I don't read any like modern political books. So I don't read the memoirs of elected officials. I don't read any, I just feel like I get enough of it in my job. So my reading list is just other things. It's history, it's narrative nonfiction, economics, et cetera. And that's my bias because I'm so overloaded with a lot of the stuff you're talking about. I haven't read any of Obama's books. I didn't read John Bolton's book or, you know, I don't read any of that stuff, </s><s>[INST] although I'm sure there is value to be gleaned from it. What about the other candidate that, according to the subreddit, and as you mentioned, you've criticized it a little bit, Marianne Williamson, do you think, [/INST] what are the pros and cons of her as a candidate? This is another area where many in my audience really are angry with me. I don't find her candidacy super interesting. I'll tell you the pros and the cons as I see them. I do think that we have elected officials in the U.S., particularly presidents, from a really narrow range of backgrounds. So it's lawyers and sometimes business people, very, very often lawyers. I think we would benefit from a much greater diversity of backgrounds. And I once said, and that would include people from education, people from the science world, people with backgrounds, maybe running nonprofits, et cetera. Now, Marianne Williamson did, I guess, at one point, run some kind of small nonprofit. And some in my audience thought that credential alone would make me fall head over heels in love with the idea of a Marianne candidacy. I've interviewed for her. It's just not for me is the way I like to say it. It's the background of the woo woo type stuff is a bit off-putting to me. I understand that someone with literal Christian Bible beliefs that also I don't like, maybe I'm more willing to accept as most of our presidents, of course, have had those views because they're otherwise more qualified. But some of the things that she says just strike me as, I just don't know. I'll give you an example. When she was on with Russell Brand, she said, there's no such thing as clinical depression. It just means someone in a clinic told you, you have depression. I don't believe that to be the case. I think we have an understanding. There's two types of depression. There's like a genetic predisposition depression. There's like a acute, something's happening in my life. Temporary depression, okay. When she was asked about it recently, she said, I didn't mean it. I was just trying to impress Russell Brand. I don't know if I'm more bothered by the things she first said or by the fact that she wanted to impress Russell Brand. But it's just like, it's just really not for me. And I agree with her on, we need to take the climate more seriously. We need to expand access to hell. I'm with all of that stuff. Now I want to say one other thing about this. Anybody who wants to run should run. I am not suggesting there should be an uncontested primary for Joe Biden. Absolutely. So you think it should be contested? Well, what I mean by contested is, so there's two parts to what we mean by contested. Will the DNC organize debates? And we'll get to that in a second, but should anybody who's on the left get out of the way because Joe Biden is president and he's running for reelection? Absolutely not. The question about should there be debates? I would like there to be debates. The DNC pretty clearly isn't going to organize them. I think if you did them, you would have to say at what polling level do you qualify? And I don't know exactly where you put that number, but I think it would be a great thing to put Joe Biden on a stage with, if you can get what, 6%, 8%? I'm not really sure what the number would be. </s><s>[INST] I'm totally all for that. Why is this set of candidates, at least from my perspective, so weak? [/INST] Do you have an understanding? Do you have an understanding? There's a lot of different answers to this. One aspect of this, which I think is more of a socio-cultural thing, which I've recently read about to some degree, is the job actually turns off the people who would be best at it because of what you need to do to become president. And it includes all but completely abandoning your existing day-to-day life job, which you may depend on, and family to some degree. It's horribly negative, as we already talked about. And at the end of all of that, you either lose and then have to rebuild, and maybe you're not in a position to be able to do that, or you win, and then now you've got four years of being one of the most hated people, no matter how good a job you do. So I think by its nature, it turns off a lot of people that would otherwise be good. I also think that there's a lot of posturing from within the parties about, well, you might be good, but it's not your time yet, so you should wait, and then let's talk about maybe a Senate seat here and there. So it's like a company, essentially, and they're figuring out where they want to place people. I think all of these things make it so we end up with candidates most people aren't super thrilled with. </s> |
<s>[INST] And in that [/INST] picture, </s><s>[INST] what would be the role of creativity. But a few people ask about creativity, do you think when you think about intelligence? I mean, certainly the story you told the beginning of your book involved, you know, creating movies and so on, sort of making making money. You know, you can make a lot of money in our modern world with music and movies. So if you are an intelligence system, you may want to get good at that. But that's not necessarily what I mean by creativity. Is it important on that complex goals where the sea is rising for there to be something creative or or am I being very human centric and thinking, creativity is somehow special relative to intelligence? [/INST] Mayan hunches a we should think your creativity simply as an aspect of intelligence. And uh we we have to be very careful with the human vanity we have we have this tendency very often want to say as soon as machines can do something, we try to diminish it and saying oh but that's not like real intelligence, you know because they're not creative or there were or this or that the other thing um if we ask ourselves to write down a definition of what we actually mean by being creative, what we mean by Andrew wiles what he did there for example, don't we often mean that someone takes very unexpected leap. It's not like taking 573 and multiplying it by 224 by just a step of straightforward cookbook like rules. Right? You can maybe make it, you make a connection between two things that people have never thought was connected, surprising or something like that. Huh? I think I think this is an aspect of intelligence and uh this is actually one of the most important aspect of it. Maybe the reason we humans tend to be better at it than traditional computers because it's something that comes more naturally if you're a neural network than if your traditional logic gates based computer machine. You know we physically have all these connections. Mhm. And that did you activate here, activated here, activate here being, you know, a my hunch is that if we ever build a machine, well, you could just give it the task. Hey, hey, uh you say, hey, you know, I just realized that I want to travel around the world instead this month. Can you teach my G. I course for me? And it's like, okay, I'll do it and it does everything that you would have done and improvises and stuff that would, in my mind involve a lot of creativity. </s><s>[INST] Yeah, So it's actually a beautiful way to put it. I think we do try to grab grasp at the, you know, the definition of intelligence is everything. We don't understand how to build. So like so we, as humans try to find things well that we have our machines don't have. And maybe creativity is just one of the things, one of the words we use to describe that, that's a really interesting way to put [/INST] it, and I don't think we need to be that defensive. I don't think anything good comes out of saying, well, where somehow special, you know, it's um contrary wise, there are many examples in history of where trying to pretend that we're somehow superior to all other intelligent beings has led the pretty bad results, right? Nazi Germany? They said that they were somehow superior to other people. Uh today we still do a lot of cruelty to animals by saying that we're so superior somehow. And they can't feel pain, slavery was justified by the same kind of just really weak arguments. And and I don't think if we actually go ahead and build artificial general intelligence can do things better than us. I don't think we should try to found our self worth on some sort of bogus claims of superiority in terms of our intelligence. I think we should instead find our calling and the meaning of life from from the experiences that we have. You know, I can have, I can have very meaningful experiences, even if there are other people who are smarter than me, you know? Okay, when I go to faculty meeting here and I were talking about something that I certainly realize, oh, but he has an old price. He has an old price. He has a surprise. I don't have, what does that make me enjoy life any less? Or I enjoy talking to those people. That's of course not, you know, and contrary wise, I I I feel very honored and privileged to get to interact with with other very intelligent beings that are better than me. A lot of stuff. So I don't think there's any reason why we can't have the same approach with intelligent machines. </s><s>[INST] That's a really interesting. So people don't often think about that. They think about when there's going if there's machines that are more intelligent, you naturally think that that's not going to be um a beneficial type of intelligence, you don't realize it could be, you know, like peers with Nobel prizes that that would be just fun to talk with, and they might be clever about certain topics and uh you can have fun having a few drinks with them, so [/INST] well also, you know, another example, we can all relate to it of why it doesn't have to be a terrible thing to be impressed with the presence of people or even smarter than us all around is when, when you and I were both two years old, I mean, our parents were much more intelligent than us, right? Worked out okay, because their goals were aligned with our goals and that I think is really the number one key issue we have to solve if we value in line of the value alignment problem. Exactly. Because people who see too many Hollywood movies with lousy science fiction, plot lines, they worry about the wrong thing, right? They worry about some machines, certainly turning evil, it's not malice, they wish that is the concern its competence. By definition intelligence makes you makes you very competent if you have a more intelligent goal. Playing, Computer playing is the less intelligent one. And when we define intelligence is the ability to accomplish, go winning right, it's going to be the more intelligent one that wins and if you have human and then you have an A g that's more intelligent than always, and they have different goals, guess who's going to get their way right? So I was just reading about, I was just reading about this uh particular rhinoceros species that was driven extinct just a few years ago, a bummer is looking at this cute picture, mommy rhinoceros with its it's child, you know, why did we humans drive it to extinction? Wasn't because we were evil rhino haters as a whole. It was just because we are goals weren't aligned with those of the rhinoceros, and it didn't work out so well for the rhinoceros because we were more intelligent, right? So I think it's just so important that if we ever do build a G i before we unleash anything, we have to make sure that it's it learns to understand our goals, adopts our goals and it retains those goals. </s> |
End of preview. Expand
in Dataset Viewer.
Intro
This dataset (5K) formats an existing podcast dataset (64bits/lex_fridman_podcast_for_llm_vicuna) for llama 3 chat model fine tuning. It represents a compilation of audio-to-text transcripts from the Lex Fridman Podcast. The Lex Fridman Podcast, hosted by AI researcher at MIT, Lex Fridman.
Problems
There might be some minor issues during the transcribe phase.
Next Step
Use whisper to directly load the podcast and transcribe it in this format.
- Downloads last month
- 48