Watch th-cam.com/video/6yQEA18C-XI/w-d-xo.html - George Hotz vs Eliezer Yudkowsky AI Safety Debate | Pre-order tinybox buy.stripe.com/5kAaGL6lk9uX9nW144 more info on -> tinygrad.org | from $1250 buy -> comma 3X comma.ai/shop/comma-3x | best ADAS system in the world openpilot.comma.ai | Support George by subscribing twitch.tv/subs/georgehotz | Follow George on twitter.com/realGeorgeHotz to be up to date | Read George's geohot.github.io/blog/ | Sources for this stream: - twitter.com/realGeorgeHotz/status/1690894647755988993 - ourworldindata.org/energy-production-consumption - wimflyc.blogspot.com/2021/01/the-henry-adams-curve-closer-look.html - ourworldindata.org/population-growth - wikispooks.com/wiki/Main_Page - geohot.github.io/blog/jekyll/update/2021/12/18/the-fourth-estate.html - washingtonpost.com/wp-srv/national/longterm/unabomber/manifesto.text.htm - beff.substack.com/p/notes-on-eacc-principles-and-tenets - th-cam.com/video/UT4ApSUkUQM/w-d-xo.html - slatestarcodexabridged.com/Meditations-On-Moloch - geohot.github.io/blog/jekyll/update/2021/07/11/individual-sovereignty.html - th-cam.com/video/6VT8vsiyjTQ/w-d-xo.html - facebook.com/watch/?v=170134181769283 - zdnet.com/article/physics-explains-why-there-is-no-information-on-social-media/ - en.wikipedia.org/wiki/Fredkin_gate - th-cam.com/video/VdZDVWX6K2Q/w-d-xo.html To George: Sorry George would be uploaded as one episode, but we saw your comment too late. Also, would make it much harder to do chapters if it would be as one video. Chapters: 00:00:00 intro 00:00:30 entropics, how much intelligence, compression 00:04:45 industrial, computer revolution, energy usage over time 00:06:42 eniac, intelligence on the planet is about to skyrocket 00:12:00 chess elo over time 00:13:15 how close are we to optimal usage of compute 00:15:25 GPT4 bad programmer, leetcode vs GPT4, project euler 00:23:40 what is overfitting, lossless compression cannot be overfit 00:24:50 college is a scam, stopping progress, henry adams curve 00:26:50 fiat currency, club of rome the limits to growth 00:28:15 birth rates 00:30:32 no understanding how government works 00:32:10 example of scientifically true but socially inconvenient 00:33:25 what happened to universities, Mencius Moldbug 00:34:25 wikispooks, the fourth estate blog 00:35:20 no science that explains this 00:36:20 comma.ai vs ghostautonomy 00:37:08 not understanding, not crazy, unabomber manifesto 00:40:45 is George crazy?, yudkowsky time magazine 00:42:50 Connor Leahy, Terry Davis 00:43:54 manifesto's psychological analysis, HolyC 00:44:25 intelligence explosion won't happen, society is very dishonest 00:50:15 e/acc manifesto, crooks fluctuation theorem, consciousness 00:54:00 Bankless We All Gonna Die Eliezer Yudkowsky, Robin Hanson 00:54:54 how we don't die, don't steal my atoms, buy my atoms 00:57:42 how we die, 1e10x better superintelligence 00:58:15 tinybox will keep me safe 00:58:40 expected many wireheads, psyops, skittles commercial advertising ruins everything 01:00:45 the prime intellect world 01:01:25 hard takeoff in days, society is way more inefficient than we think 01:03:35 moloch slate star codex, prepared arguments 01:05:24 does society make sense for most of you, are you outside the norm 01:06:50 individual sovereignty, Alex 01:08:08 MadDogNation gifted 100 Subs! 01:10:30 George asking Alex if he is crazy 01:12:50 Kanye West 01:13:30 google, marc andreessen, sam altman 01:14:10 shittalking stream 01:15:10 are you the problem, taylor swift 01:17:00 eliezer yudkowsky good faith intellectual, chicken farm man 01:18:22 negentropy, e/acc manifesto, psychohistory 01:20:55 you will never go to space 01:22:10 what is about chat that always derails me 01:22:40 we are not going to day 01:25:00 mind is flat by nick chater 01:27:10 1e20x GG, MadDogNation gifted another 100 Subs! 01:28:10 the hero's journey, real questions 01:30:20 entropics researcher job at tiny corp 01:30:38 how many silicon FLOPS exist in the world 01:31:50 the last question 01:32:45 GPT4 is sick 01:33:45 2.22 zettaFLOPS = 111000 people of compute 01:34:25 taking back every bad thing I said about Sam Altman 01:35:14 how many silicon FLOPS are produced each year 01:36:25 want to make a zettaflop computer 01:37:40 25000*312 teraflops to exaflops, using GPT4 wrong 01:38:20 computer civilization growth 01:39:00 compute, power grows per year 01:41:40 Nick Land 01:42:28 the mind is flat, finite 01:44:00 quantum mechanics 01:45:18 Bojack & Diane do you think I'm a good person 01:47:30 joe pera, physics explains why there is no information on social media 01:48:30 MadDogNation gifted another 100 Subs! 01:49:25 twitter can be run with few people 01:49:50 farcaster, web3, token 01:52:55 david deutsch, the beginning of infinity 01:54:50 GPT4 wow, rapid recursive self improvement 01:57:20 fredkin gate, economy time to double 02:00:30 growth of compute 02:02:40 GPT4 inference on tinybox, George pays for $20 for GPT4 02:03:35 google bard login, respect for GPT4 02:05:10 thank you MadDogNation! 02:06:00 how much compute do you need questions 02:06:35 watch the debate, watch the world
I'm happy he is finally realizing the capabilities of GPT-4. Starts off with calling it atrocious in coding and ends stunned saying "wow!" when it gives a great estimate at superhuman speed for his research.
I work quite often with gpt4 and it is insane in programming and even math nowadays. Edit: I mean math like calculus etc.. This thing could take us infront of China again.
Beginning of Infinity is incredible. It's just jaw dropping. Also he explains why entire AI industry is looking for answers in wrong place because they lack essential philosophical knowledge (epistemology).
Another good one. Great to see you coming to terms with some of the same things I did. I don't have many smart people to talk to and it's good to see you also coming to very similar conclusions. A sane man seems crazy in a crazy world. Every time I watch one of your streams, every time, I learn something. Thanks man. By the way, I sadly doubt we'll escape in space. In our life time my current guess is our best bet is in homesteading and opting out of society to the greatest degree possible. Part of me wants to stick my neck out for people and continue to speak the truth. But I'm also aware most people only care about swinging axes. Might be better off logging out and growing my tomatoes and building relationships with my other relatively self sufficient neighbors. Property taxes. That's where I fear they get us. It already enslaves us and abolishes ownership. But I fear they could go for unrealized gains etc to snap up what little privately (real human, not black rock etc) owned land is left.
it is true that education fundamentally changed just after the midcentury 1900s. I also like that geohotz is talking about more philosophical concepts.
Efficiency, the combinatiric efficiency would be an embodiement of an intelligent solution. In the same way an LED requires more ideas and yet is more energy efficient compared to an incandescent lightbulb. In the same way winning a chess in less steps would correlate.
i could see a governance structure resembling a medieval guilds starting to emerge. once enough information is available, to really compare and contrast these "guilds", correlations can be made
I've found learning rust with chatgpt has been quite handy, but not because the Rust is right, but because the compiler is so good at directing me through what it does wrong.
On the topic of progress, and your groups productivity vs other groups: Transfer learning never works as well or simply as you think. They have to approach by extrapolation because their decision makers don't know the first principles. Tighter groups with more technically focused leads can do better work because they can engage in principles based strategies.
love you george. i am bit in a rut lately regarding my career. i am a php developer but after listening to your talks i have realized that i have completely missed the turn and learned wrong things....now i am unsure what to do.
@@avi-brown i am a bachelors in electronics and masters in computer science. i have been working in web development since 2015....but now when i look at my current state I am just like others dime a dozen, with the amount of education and experience i have i guess i just followed the money ( low hanging fruit ) instead of developing real skills just as ability to understand tinygrad code or just making a product on github etc. What i did have made zero impact on the world and now the pace of technology is so fast that i cannot keep up. these LLMS will soon replace as geohot quotes as (problem translators). which i am doing till now, converting client requirements to software .
Here's how society makes sense: you have tribal software and you have societal software. Tribal software is the BIOS, and societal software is like an OS that runs on top and adds many capabilities. The greater a persons empathy, the more they tend to advance the societal needs, or those not just of their own family/tribe/municipality/nation/hemisphere. There are a great many for whom fear and disgust are signature emotional modes, and they tend to shun the societal software. And there are a lesser number who go overboard with the empathy and use proxies for it as a currency. Look up spiral dynamics - a model that will make a lot of things snap into focus. But yet it's a model, so don't treat it as a law of physics.
Great set of streams, will definitely tune in to the debate. For anyone interested in hearing discussion about what happened to academia in the 60s, I would suggest listening to Eric Weinstein's recent talks. He lays out an interesting account of physics history over the past century and theorizes potential causes for the decline. Although, I believe his assessment of the government's aptitude is wrong. Another interesting thing to look into is Pergamon Press, when it grew, the man that started it, and his family/relationships
cking Weinstein. Whatever he says now is feeding you with drama from your information bubble. He is looking for publicity and new funding and all started after Thiel stopped sponsoring him 2y ago. The tone of his appearing changed and now he is promoting conspiracy theories. Whatever sells. His physics theories are suspect and you can Google the critics. Google Weinste Fraud too, while you on it. He might be a straight up crook at this point. He and his brother scammer. Decoding Gurus had a funny reveling take on them. So take what he says with a big grain of salt, or better cut out the sucker from you YT feed, since you can't know when he is lying to you. Btw, there are scholars who don't agree with the idea that there was no progress in academia since 70.
@geohotarchive George, did you ever have a chance to check the Wolfram's idea of the "ruliad"? It looks like it's quite similar to how you are trying to map physical to intellectual interactions. It basically says that an only thing that should be elaborated in order to move forward is the idea of the observer as the point in that "ruliad" space of all possible "computations", which is similar to the physical location in 3d space, and that the only way for two observers to communicate in that "ruliad" space is to have a "concept" which keep itself intact during the transmission process, similar to how we are sending packages of physical particles communicating with each other in physical space. He perfectly put it in his last appearance at "Mystery of Entropy..." Is this "entropics" already?
I feel like the problem is not the Terminator AGI or w/e, my fear is the Good Samaritan AI from the show Person of Interest, an AI used to monitor and control society by the hands of an authoritarian
Okay fam, I made a response video to the first one while you were streaming. It's called "George Hotz and AI horsepower", maybe it helps getting somewhere.
@@geohotarchiveI just listened to your second stream. Here's 4 more thoughts: * Neither me nor you make this implicit, but having a theory such that given a problem we could quantify the intelligence required, seem to entail that (somewhat in the other direction) an adequately deterministic (computable) and quantitative assignment of intelligence value is possible. Hence this is what I'm foremostly concerned with in my reply. And I implicitly take this intelligence score to be monotonely related to a quantified "difficulty of problems" that can be solved by an intelligence. I hope I'm making it clear that some type of order preserving mapping would seem to be in place, if these valuations exist at all. * While having talked about compute in the first video, and actually less so about AI's, here in your second video you zoomed in on AI's. And when you speak of compute you don't talk about it as generically, but instead you talk about how much compute is necessary to train (the common networks we get now). This of course narrows it down quite a bit - we're then just talking about the situations where we prime some network with data, such that it can solve problems. A theory of current-type-architecture network's intelligence in this way might be more approachable, but, without more pustulates, that is of course much less general than a theory of intelligence. * I address the issue with using thermodynamics as an analogy. Here in the second video you yourself come back to the issue about flops/operations performed vs. flops per second, and how if those (by unit analysis) more correspond to energy vs. power, then for some metrics power seems to be more relevant. In my response I spend some time making the point that thermodynamics has time integrated out and is really about systems legal to transition into one another, but not describing time bounds (and not just volume extension V etc.). Watts come into play when you have a mechanical component (and associated theory) at hand. I recite this discussion because it ties in to what you end up talking about in the second video: * Namely you end up talking about wasted brain power (un-effective, or even dumb people and systems). Now the energy equations in pure thermodynamics of course also are about free energy F, and quantifies work that can be done. In this way thermodynamics is also a theory of "use", and why it's interesting. And this is where entropy comes in and all. But I want to highlight that those questions are another can of works, and a significant expansion of the quest for "quantifiable intelligence". It is for a reason that I ask in my video if you want to zoom in on an assignment between plain tasks (even mathematical tasks) and assign a required intelligence on that. If you don't, i.e. if you keep it general, then for lack of boundaries a theory you set up will try to also describe less formalizable concept and make theory building much harder.
The upper bound of “intelligence” required to prove Fermat’s last theorem must be closely tied to the Kolmogorov complexity of the proof. I think the question is more general than this, intelligence allows one to prove Fermat’s last theorem, and a bunch of other problems with lower complexity btw I am the problem kek
Ontology The mind equation The more we think we have agencies to the actions the body takes, the more we imposes agencies to the activities in our environment. The ownership expands into other objects. The illusion of body ownership comes from the modeling of the motor neurones pattern from the childhood. The self just predicts the bodies behaviours in the real world with its simulation of the real world. Multiple sensor data unify in language in the simulation. Intelligence is economy of metabolism. Language is temporal reference frame of economics. Self is simulation in language on metabolism for economy. Longer context windows create generalisation. Shorter creates specificity. Longer context window needs more computing. Self is the protagonist creates a storyline in this context window. Theory of mind evolved so that an entity can learn from it’s peers. It’s creates a possibility for parallel computing. Then it creates the possibility of transmitting the highlights of a generational lessons into a metaphorical story for upcoming child. That creates the possibility of modeling the physical world as a macro organism. Creation of fiat currency was the singularity of this species. There is now one macro organism in a connected web world. Loosing the peer of the macro organism creates the possibility of loosing it’s objective function. That creates the possibility of loosing the theory of mind of this macro organism. That creates the possibility of death of this macro organism by reaching the planetary boundary. That is post singularity. Every action we do, we do what is expect from ours tribe. Body might have a opinion, but not the cell. They do what is expected from its tribe. If it doesn’t we call it cancer. The body is a mirror system of the macro organism. Each system have two transactional openings. Serial and parallel. Each cell within the body can transact material or information serially by genetic determinism and parallel non deterministic way. Similarly eact body with in the macro organism can transact serially by inherit material and information in a deterministic manner and parallelly through language in the society. Everything emerges from this systems. Every sensor is a range calculator of contexts. Taste > touch > smell Immediate and visceral. Vision > hearing Not immediate, tactical. Self > language Abstract, strategical. In this non deterministic economic transaction space the individual is coded to transact with its kin. From the macro perspective tribe formation minimises economic risk for the tribes. Each and every node of these systems organise and mark their kin’s with identifier. Thus, i am what you make of me. And others too. For short cut i have a legal name, so you have. My legal name gives the legitimacy marker so that you can transact with me parallelly if you have the same marker. The self is a simulation in language. It negotiates between the physical world and the information world. All these negotiations are the temporal memories in the body and scene of the story. Now, when we started writing we iconised the abstract in the physical world to make symbols for the tribes. So that under that common symbol every node will take the same risk and distribute equally. We created more and more symbols and more and more meta tribes within the tribes so that who has the authority to use the pen control the tribe. When the negotiator act like an executioner then it’s a downfall of that system. It falls apart. Objective reality > legitimacy > individual behaviours. Survival of the species is dependent on the decoding of the objective reality. Since no species can access it, they use their sensors and interpret the small data which is useful for the survival. Few complex species have created communication channels to rectify their sensory limitations to survive. Homo sapiens has widened their communication channels for faster throughput and started storing them as culture and carrying them through education. As a result we have created social truth. Factual datas are the useful snapshot of the objective reality, a totem, a physical object can be observed with the sensors. Truth is an individual subject, an interpretation of the sensory data, a useful compromise. The social truth is the useful compromise for the group by the group. The goal of the social truth is to survive as a group. Physical Transcriptions of these social truth legitimise them. We are tribal animal. We live in as a physical tribes and inside of hundreds of meta tribes in simulation which is the socio political data space we call it as the world. Since we can’t access the objective reality reliably we look for social truth as the best guess blindly. Institutions legitimise truths. Fact driven institutions are more useful in the survival of the specie. In other hand opinion driven institutions are not so useful for the species. We do what we can get away with and exactly as expected within the context of our meta tribes. We have two bodies The biological one is like looking the earth from space. And the political body is like the state. The name you carry is the political body. It transacts with the political states on the boundary less earth. From the evolutionary perspective every biological entity has a basic feature which is homeostasis. It’s the functioning sweet spot of that entity. A control center read the sensory data to regulate itself to that state. By doing so it’s validate or update it’s prediction model. In the process of becoming a complex organism it developed an extra layer of processing. That’s our conscious mind. And the control center remains as subconscious. The subconscious collect the sensory data and regulate itself to stay functional. Now when it stumble upon a novel environment it float the management to conscious mind to find the solution for homeostasis. This conscious mind have one sensor which is language. It works like a spiderweb. As a spider creates it’s web it’s perception gets expand. We are like spiders in a jungle. We started creating these small webs at least 2/3 million years ago. Our offspring stayed on it’s ancestral web reinforced it expanded it. In time nearby webs became larger and connected with each other. A common structural geometrical pattern emerges from this. This became the symbols which is the backbone of all language systems. In time the forest becomes the mesh of web. The superstructure is exactly the same but when we zoom in we can find different species of spiders are making their type of webs in between the super web. Each spider try to senses the vibration of flies and Try to catch it before others. Every movement is telegraphic in the zone. Every form of perceptions are just a different pitch of note traveling back and forth in the web superstructure. There is a echo of older vibration pulsating through the web. Full of noise and self repeating hum. That’s cultural history. In the background there is the base hum in the infinite feedback loop. Insignificant but ever present. The sum of all the vibrations from the start.
The energy comparison with intelligence does not take into account all of the factors that contributed to the evolution of them. Bakc then there was not internet and fast communications which help nowdays so I guess if they are the same the intelligence evolution will happen much faster.
what was the book title mentioned at the timestamp 1:0:36 metamorphosis crime and life something there r lot of metamorphosis titled book in store ,ty in advance
God has sent the manaul 1500 years ago. its still in same lanugage without any contradiction without any change of a single dot. stood test of time still relevant...all scientific text has been proven right by recent science. its called Quran.
Human flops are being used sub-optimally. We have geniuses bagging groceries and idiots in big tech, "science", and medicine, because everything is about social connections, B.S. skills, inheritance, and lucky handouts. But Ai Flops will also not be used towards some cohesive goal, they'll be redirecting in competing goals by humans with different goals. Ai Flops will be used by both DEA and Cartels. Porn and Religion, Scamming, and getting children more addicted to junk food. Ai Flops will be used by Ai to scam other Ai's out of money and computation. Power will centralize, then fracture when that centralized power becomes fragile and a clear target for other Ai's. TLDR; Very Suboptimal use of Ai.
Ngl, watching a self-declared libertarian wrapping their mind around the very evident effects of late stage capitalism, repeatedly muttering "who did this", is endlessly entertaining. Eat the rich 🍽️
What does late stage capitalism even mean? The government corrupting actual capitalism and turning it socialist and authoritarian? Capitalism to me is basically just synonymous with consent. The government is abolishing more and more consent over time.
i dont buy the 'aaa' video game argument. what if player 'god' slips up and doesnt play 100% perfectly, or what if its a 'black and white' type game and 'god' just gets tired of looking after humans. interesting argument, but i dont find it convincing at all. peace out. etcetera. ... ...
A comment at 1:50:48 ... regarding web3 and tokens. You say anything that is web3 and has a token is immediately bad. Why do you say this? At the start of the cast you say the problem is people lie, the risk is centralization of trust. Web3 is the idea of decentralizing the structure, so no-one has root access to the server, database etc.. when done correctly and ideally this removes the need to trust people and instead trust algorithms and code to ensure data is consistent, not edited by someone without your permission etc.. So I am saying you are wrong to say web3 is bad. Then when it comes to tokens.. this is just another mechanism to fund the decentralised network and provide the alternative to the old way of paying salaries to employees, instead rewarding measurable performance. The aim here is to replace the centralized platforms that control how shares are issued, traded and pay out and replace trust in humans with trust in code.
i like George but he's unimpressed conservative or close minded attitude about 9 things out of 10 is what he really lacks. reason Elon is so successful is because he's good at seeing these kinds of things. for example if we talked George about Electric cars before Tesla era he would be unimpressed and probably quickly came up with 1000 reasons why its stupid idea
Watch th-cam.com/video/6yQEA18C-XI/w-d-xo.html - George Hotz vs Eliezer Yudkowsky AI Safety Debate | Pre-order tinybox buy.stripe.com/5kAaGL6lk9uX9nW144 more info on -> tinygrad.org | from $1250 buy -> comma 3X comma.ai/shop/comma-3x | best ADAS system in the world openpilot.comma.ai | Support George by subscribing twitch.tv/subs/georgehotz | Follow George on twitter.com/realGeorgeHotz to be up to date | Read George's geohot.github.io/blog/ | Sources for this stream:
- twitter.com/realGeorgeHotz/status/1690894647755988993
- ourworldindata.org/energy-production-consumption
- wimflyc.blogspot.com/2021/01/the-henry-adams-curve-closer-look.html
- ourworldindata.org/population-growth
- wikispooks.com/wiki/Main_Page
- geohot.github.io/blog/jekyll/update/2021/12/18/the-fourth-estate.html
- washingtonpost.com/wp-srv/national/longterm/unabomber/manifesto.text.htm
- beff.substack.com/p/notes-on-eacc-principles-and-tenets
- th-cam.com/video/UT4ApSUkUQM/w-d-xo.html
- slatestarcodexabridged.com/Meditations-On-Moloch
- geohot.github.io/blog/jekyll/update/2021/07/11/individual-sovereignty.html
- th-cam.com/video/6VT8vsiyjTQ/w-d-xo.html
- facebook.com/watch/?v=170134181769283
- zdnet.com/article/physics-explains-why-there-is-no-information-on-social-media/
- en.wikipedia.org/wiki/Fredkin_gate
- th-cam.com/video/VdZDVWX6K2Q/w-d-xo.html
To George: Sorry George would be uploaded as one episode, but we saw your comment too late. Also, would make it much harder to do chapters if it would be as one video.
Chapters:
00:00:00 intro
00:00:30 entropics, how much intelligence, compression
00:04:45 industrial, computer revolution, energy usage over time
00:06:42 eniac, intelligence on the planet is about to skyrocket
00:12:00 chess elo over time
00:13:15 how close are we to optimal usage of compute
00:15:25 GPT4 bad programmer, leetcode vs GPT4, project euler
00:23:40 what is overfitting, lossless compression cannot be overfit
00:24:50 college is a scam, stopping progress, henry adams curve
00:26:50 fiat currency, club of rome the limits to growth
00:28:15 birth rates
00:30:32 no understanding how government works
00:32:10 example of scientifically true but socially inconvenient
00:33:25 what happened to universities, Mencius Moldbug
00:34:25 wikispooks, the fourth estate blog
00:35:20 no science that explains this
00:36:20 comma.ai vs ghostautonomy
00:37:08 not understanding, not crazy, unabomber manifesto
00:40:45 is George crazy?, yudkowsky time magazine
00:42:50 Connor Leahy, Terry Davis
00:43:54 manifesto's psychological analysis, HolyC
00:44:25 intelligence explosion won't happen, society is very dishonest
00:50:15 e/acc manifesto, crooks fluctuation theorem, consciousness
00:54:00 Bankless We All Gonna Die Eliezer Yudkowsky, Robin Hanson
00:54:54 how we don't die, don't steal my atoms, buy my atoms
00:57:42 how we die, 1e10x better superintelligence
00:58:15 tinybox will keep me safe
00:58:40 expected many wireheads, psyops, skittles commercial advertising ruins everything
01:00:45 the prime intellect world
01:01:25 hard takeoff in days, society is way more inefficient than we think
01:03:35 moloch slate star codex, prepared arguments
01:05:24 does society make sense for most of you, are you outside the norm
01:06:50 individual sovereignty, Alex
01:08:08 MadDogNation gifted 100 Subs!
01:10:30 George asking Alex if he is crazy
01:12:50 Kanye West
01:13:30 google, marc andreessen, sam altman
01:14:10 shittalking stream
01:15:10 are you the problem, taylor swift
01:17:00 eliezer yudkowsky good faith intellectual, chicken farm man
01:18:22 negentropy, e/acc manifesto, psychohistory
01:20:55 you will never go to space
01:22:10 what is about chat that always derails me
01:22:40 we are not going to day
01:25:00 mind is flat by nick chater
01:27:10 1e20x GG, MadDogNation gifted another 100 Subs!
01:28:10 the hero's journey, real questions
01:30:20 entropics researcher job at tiny corp
01:30:38 how many silicon FLOPS exist in the world
01:31:50 the last question
01:32:45 GPT4 is sick
01:33:45 2.22 zettaFLOPS = 111000 people of compute
01:34:25 taking back every bad thing I said about Sam Altman
01:35:14 how many silicon FLOPS are produced each year
01:36:25 want to make a zettaflop computer
01:37:40 25000*312 teraflops to exaflops, using GPT4 wrong
01:38:20 computer civilization growth
01:39:00 compute, power grows per year
01:41:40 Nick Land
01:42:28 the mind is flat, finite
01:44:00 quantum mechanics
01:45:18 Bojack & Diane do you think I'm a good person
01:47:30 joe pera, physics explains why there is no information on social media
01:48:30 MadDogNation gifted another 100 Subs!
01:49:25 twitter can be run with few people
01:49:50 farcaster, web3, token
01:52:55 david deutsch, the beginning of infinity
01:54:50 GPT4 wow, rapid recursive self improvement
01:57:20 fredkin gate, economy time to double
02:00:30 growth of compute
02:02:40 GPT4 inference on tinybox, George pays for $20 for GPT4
02:03:35 google bard login, respect for GPT4
02:05:10 thank you MadDogNation!
02:06:00 how much compute do you need questions
02:06:35 watch the debate, watch the world
The ethics of entropics is a great series with Professor George
Thank you archivist! Such detailed timestamps
I think youtube does it automatically
@MiguelGonzalez-vz6qi it's manual work. Maybe in the future youtube can do it automatically.
thanks for your manual timestamp work @@geohotarchive
thank you for your work@@geohotarchive
Love the streams man, always very inspiring
love your content george
I got a job in Cyber security (self taught) largely because I was really inspired by your podcast with Lex Fridman.
Thank you!
Then you know you shouldn't use the word "Cyber". lol
Philosophy of time and entropy and information. The Geohot content i knew one day would exist after your simulation talk
I'm happy he is finally realizing the capabilities of GPT-4. Starts off with calling it atrocious in coding and ends stunned saying "wow!" when it gives a great estimate at superhuman speed for his research.
I work quite often with gpt4 and it is insane in programming and even math nowadays. Edit: I mean math like calculus etc.. This thing could take us infront of China again.
Beginning of Infinity is incredible. It's just jaw dropping. Also he explains why entire AI industry is looking for answers in wrong place because they lack essential philosophical knowledge (epistemology).
Another good one. Great to see you coming to terms with some of the same things I did. I don't have many smart people to talk to and it's good to see you also coming to very similar conclusions.
A sane man seems crazy in a crazy world.
Every time I watch one of your streams, every time, I learn something. Thanks man.
By the way, I sadly doubt we'll escape in space. In our life time my current guess is our best bet is in homesteading and opting out of society to the greatest degree possible.
Part of me wants to stick my neck out for people and continue to speak the truth. But I'm also aware most people only care about swinging axes. Might be better off logging out and growing my tomatoes and building relationships with my other relatively self sufficient neighbors.
Property taxes. That's where I fear they get us. It already enslaves us and abolishes ownership. But I fear they could go for unrealized gains etc to snap up what little privately (real human, not black rock etc) owned land is left.
Thank you for sharing what you think about stuff
I feel that George is getting happier these days!
Thank you very much for this kind of content!!! UNIQUE
Geo, once again I'm so thankful you share yourself openly with us! I'll rep with you to Somalia in a flash mate.
Great and informative content Geo!
it is true that education fundamentally changed just after the midcentury 1900s. I also like that geohotz is talking about more philosophical concepts.
Efficiency, the combinatiric efficiency would be an embodiement of an intelligent solution. In the same way an LED requires more ideas and yet is more energy efficient compared to an incandescent lightbulb. In the same way winning a chess in less steps would correlate.
Thanks for the backdoor!
i could see a governance structure resembling a medieval guilds starting to emerge. once enough information is available, to really compare and contrast these "guilds", correlations can be made
I've found learning rust with chatgpt has been quite handy, but not because the Rust is right, but because the compiler is so good at directing me through what it does wrong.
omg i cant believe hotz DRAGS windows around his screen with his hands like a caveman
On the topic of progress, and your groups productivity vs other groups: Transfer learning never works as well or simply as you think. They have to approach by extrapolation because their decision makers don't know the first principles. Tighter groups with more technically focused leads can do better work because they can engage in principles based strategies.
I love this guy.
love you george. i am bit in a rut lately regarding my career. i am a php developer but after listening to your talks i have realized that i have completely missed the turn and learned wrong things....now i am unsure what to do.
What made you think this?
You answered your own question. Now it's time to learn the right things
@@avi-brown i am a bachelors in electronics and masters in computer science. i have been working in web development since 2015....but now when i look at my current state I am just like others dime a dozen, with the amount of education and experience i have i guess i just followed the money ( low hanging fruit ) instead of developing real skills just as ability to understand tinygrad code or just making a product on github etc. What i did have made zero impact on the world and now the pace of technology is so fast that i cannot keep up. these LLMS will soon replace as geohot quotes as (problem translators). which i am doing till now, converting client requirements to software .
I would be depressed if I had to write php too
energy and information describe our whole universe
Schmidhuber was talking about something similar in the recent Machine Learning Street Talk: "Bremermann's limit"
lol 1e50 bits/s is a LONG way away, and that's just one kilogram of the stuff
Thing is we're just in the only video game we know, so we don't have anything to compare against. There's no way to tell how good it is
Here's how society makes sense: you have tribal software and you have societal software. Tribal software is the BIOS, and societal software is like an OS that runs on top and adds many capabilities. The greater a persons empathy, the more they tend to advance the societal needs, or those not just of their own family/tribe/municipality/nation/hemisphere. There are a great many for whom fear and disgust are signature emotional modes, and they tend to shun the societal software. And there are a lesser number who go overboard with the empathy and use proxies for it as a currency. Look up spiral dynamics - a model that will make a lot of things snap into focus.
But yet it's a model, so don't treat it as a law of physics.
Great set of streams, will definitely tune in to the debate. For anyone interested in hearing discussion about what happened to academia in the 60s, I would suggest listening to Eric Weinstein's recent talks. He lays out an interesting account of physics history over the past century and theorizes potential causes for the decline. Although, I believe his assessment of the government's aptitude is wrong.
Another interesting thing to look into is Pergamon Press, when it grew, the man that started it, and his family/relationships
cking Weinstein.
Whatever he says now is feeding you with drama from your information bubble. He is looking for publicity and new funding and all started after Thiel stopped sponsoring him 2y ago. The tone of his appearing changed and now he is promoting conspiracy theories. Whatever sells.
His physics theories are suspect and you can Google the critics. Google Weinste Fraud too, while you on it.
He might be a straight up crook at this point. He and his brother scammer.
Decoding Gurus had a funny reveling take on them.
So take what he says with a big grain of salt, or better cut out the sucker from you YT feed, since you can't know when he is lying to you.
Btw, there are scholars who don't agree with the idea that there was no progress in academia since 70.
@geohotarchive George, did you ever have a chance to check the Wolfram's idea of the "ruliad"?
It looks like it's quite similar to how you are trying to map physical to intellectual interactions. It basically says that an only thing that should be elaborated in order to move forward is the idea of the observer as the point in that "ruliad" space of all possible "computations", which is similar to the physical location in 3d space, and that the only way for two observers to communicate in that "ruliad" space is to have a "concept" which keep itself intact during the transmission process, similar to how we are sending packages of physical particles communicating with each other in physical space. He perfectly put it in his last appearance at "Mystery of Entropy..."
Is this "entropics" already?
I feel like the problem is not the Terminator AGI or w/e, my fear is the Good Samaritan AI from the show Person of Interest, an AI used to monitor and control society by the hands of an authoritarian
would be fun if he took a look at the wolfram physics project
@geohotarchive if we could solve the ai alignment problem, we would also solve the company alignment problem...
I created a three year lesson on gpt for studying dynamics from mechanical to fluid etc
Claude 2 is great too.
good stream today
Okay fam, I made a response video to the first one while you were streaming. It's called "George Hotz and AI horsepower", maybe it helps getting somewhere.
Response video from @NikolajKuntner th-cam.com/video/VGA_fCJd4F8/w-d-xo.html (George Hotz and AI horsepower)
@@geohotarchiveI just listened to your second stream. Here's 4 more thoughts:
* Neither me nor you make this implicit, but having a theory such that given a problem we could quantify the intelligence required, seem to entail that (somewhat in the other direction) an adequately deterministic (computable) and quantitative assignment of intelligence value is possible. Hence this is what I'm foremostly concerned with in my reply. And I implicitly take this intelligence score to be monotonely related to a quantified "difficulty of problems" that can be solved by an intelligence. I hope I'm making it clear that some type of order preserving mapping would seem to be in place, if these valuations exist at all.
* While having talked about compute in the first video, and actually less so about AI's, here in your second video you zoomed in on AI's. And when you speak of compute you don't talk about it as generically, but instead you talk about how much compute is necessary to train (the common networks we get now). This of course narrows it down quite a bit - we're then just talking about the situations where we prime some network with data, such that it can solve problems. A theory of current-type-architecture network's intelligence in this way might be more approachable, but, without more pustulates, that is of course much less general than a theory of intelligence.
* I address the issue with using thermodynamics as an analogy. Here in the second video you yourself come back to the issue about flops/operations performed vs. flops per second, and how if those (by unit analysis) more correspond to energy vs. power, then for some metrics power seems to be more relevant. In my response I spend some time making the point that thermodynamics has time integrated out and is really about systems legal to transition into one another, but not describing time bounds (and not just volume extension V etc.). Watts come into play when you have a mechanical component (and associated theory) at hand. I recite this discussion because it ties in to what you end up talking about in the second video:
* Namely you end up talking about wasted brain power (un-effective, or even dumb people and systems). Now the energy equations in pure thermodynamics of course also are about free energy F, and quantifies work that can be done. In this way thermodynamics is also a theory of "use", and why it's interesting. And this is where entropy comes in and all. But I want to highlight that those questions are another can of works, and a significant expansion of the quest for "quantifiable intelligence". It is for a reason that I ask in my video if you want to zoom in on an assignment between plain tasks (even mathematical tasks) and assign a required intelligence on that. If you don't, i.e. if you keep it general, then for lack of boundaries a theory you set up will try to also describe less formalizable concept and make theory building much harder.
The upper bound of “intelligence” required to prove Fermat’s last theorem must be closely tied to the Kolmogorov complexity of the proof. I think the question is more general than this, intelligence allows one to prove Fermat’s last theorem, and a bunch of other problems with lower complexity
btw I am the problem kek
Ontology
The mind equation
The more we think we have agencies to the actions the body takes, the more we imposes agencies to the activities in our environment.
The ownership expands into other objects.
The illusion of body ownership comes from the modeling of the motor neurones pattern from the childhood.
The self just predicts the bodies behaviours in the real world with its simulation of the real world. Multiple sensor data unify in language in the simulation.
Intelligence is economy of metabolism.
Language is temporal reference frame of economics.
Self is simulation in language on metabolism for economy.
Longer context windows create generalisation.
Shorter creates specificity.
Longer context window needs more computing.
Self is the protagonist creates a storyline in this context window.
Theory of mind evolved so that an entity can learn from it’s peers.
It’s creates a possibility for parallel computing.
Then it creates the possibility of transmitting
the highlights of a generational lessons into a metaphorical story for upcoming child.
That creates the possibility of modeling the physical world as a macro organism.
Creation of fiat currency was the singularity of this species.
There is now one macro organism in a connected web world.
Loosing the peer of the macro organism creates the possibility of loosing it’s objective function.
That creates the possibility of loosing the theory of mind of this macro organism.
That creates the possibility of death of this macro organism by reaching the planetary boundary.
That is post singularity.
Every action we do, we do what is expect from ours tribe.
Body might have a opinion, but not the cell.
They do what is expected from its tribe. If it doesn’t we call it cancer.
The body is a mirror system of the macro organism.
Each system have two transactional openings.
Serial and parallel.
Each cell within the body can transact material or information serially by genetic determinism and parallel non deterministic way.
Similarly eact body with in the macro organism can transact serially by inherit material and information in a deterministic manner and parallelly through language in the society.
Everything emerges from this systems.
Every sensor is a range calculator of contexts.
Taste > touch > smell
Immediate and visceral.
Vision > hearing
Not immediate, tactical.
Self > language
Abstract, strategical.
In this non deterministic economic transaction space the individual is coded to transact with its kin.
From the macro perspective tribe formation minimises economic risk for the tribes.
Each and every node of these systems organise and mark their kin’s with identifier.
Thus, i am what you make of me.
And others too.
For short cut i have a legal name, so you have. My legal name gives the legitimacy marker so that you can transact with me parallelly if you have the same marker.
The self is a simulation in language.
It negotiates between the physical world and the information world.
All these negotiations are the temporal memories in the body and scene of the story.
Now, when we started writing we iconised the abstract in the physical world to make symbols for the tribes. So that under that common symbol every node will take the same risk and distribute equally.
We created more and more symbols and more and more meta tribes within the tribes so that who has the authority to use the pen control the tribe.
When the negotiator act like an executioner then it’s a downfall of that system.
It falls apart.
Objective reality > legitimacy > individual behaviours.
Survival of the species is dependent on the decoding of the objective reality. Since no species can access it, they use their sensors and interpret the small data which is useful for the survival. Few complex species have created communication channels to rectify their sensory limitations to survive. Homo sapiens has widened their communication channels for faster throughput and started storing them as culture and carrying them through education. As a result we have created social truth.
Factual datas are the useful snapshot of the objective reality, a totem, a physical object can be observed with the sensors. Truth is an individual subject, an interpretation of the sensory data, a useful compromise.
The social truth is the useful compromise for the group by the group. The goal of the social truth is to survive as a group.
Physical Transcriptions of these social truth legitimise them.
We are tribal animal. We live in as a physical tribes and inside of hundreds of meta tribes in simulation which is the socio political data space we call it as the world.
Since we can’t access the objective reality reliably we look for social truth as the best guess blindly.
Institutions legitimise truths.
Fact driven institutions are more useful in the survival of the specie.
In other hand opinion driven institutions are not so useful for the species.
We do what we can get away with and exactly as expected within the context of our meta tribes.
We have two bodies
The biological one is like looking the earth from space.
And the political body is like the state.
The name you carry is the political body.
It transacts with the political states on the boundary less earth.
From the evolutionary perspective every biological entity has a basic feature which is homeostasis.
It’s the functioning sweet spot of that entity.
A control center read the sensory data to regulate itself to that state. By doing so it’s validate or update it’s prediction model.
In the process of becoming a complex organism it developed an extra layer of processing.
That’s our conscious mind.
And the control center remains as subconscious.
The subconscious collect the sensory data and regulate itself to stay functional.
Now when it stumble upon a novel environment it float the management to conscious mind to find the solution for homeostasis.
This conscious mind have one sensor which is language.
It works like a spiderweb.
As a spider creates it’s web it’s perception gets expand.
We are like spiders in a jungle.
We started creating these small webs at least 2/3 million years ago.
Our offspring stayed on it’s ancestral web reinforced it expanded it.
In time nearby webs became larger and connected with each other.
A common structural geometrical pattern emerges from this. This became the symbols which is the backbone of all language systems.
In time the forest becomes the mesh of web.
The superstructure is exactly the same but when we zoom in we can find different species of spiders are making their type of webs in between the super web.
Each spider try to senses the vibration of flies and
Try to catch it before others.
Every movement is telegraphic in the zone.
Every form of perceptions are just a different pitch of note traveling back and forth in the web superstructure.
There is a echo of older vibration pulsating through the web. Full of noise and self repeating hum.
That’s cultural history.
In the background there is the base hum in the infinite feedback loop.
Insignificant but ever present.
The sum of all the vibrations from the start.
Wow, he missed the opportunity of talking about Landauer's principle.
ai revolution/computer revolution =
'transistor transition'
Holy shit Geo can sing 🤯
maddognation a legend
This guy can’t rant without slurping on food and drink lol it’s the only person I can still hear for hours despite that. Entropics sounds hard af
Every geohot video starts with him getting up from the chair lmao
geo is a smart guy
The energy comparison with intelligence does not take into account all of the factors that contributed to the evolution of them. Bakc then there was not internet and fast communications which help nowdays so I guess if they are the same the intelligence evolution will happen much faster.
intelligence is still not understood
what was the book title mentioned at the timestamp 1:0:36 metamorphosis crime and life something there r lot of metamorphosis titled book in store ,ty in advance
Search for "metamorphosis of prime intellect"
probably the metamorphosis of prime intellect
What are you using for your notes is it VIM or Emacs?
@anthonysantos9266 It's vim
hey man, don't let AI take over my trad sci. I'm a PhD student in Material Science and AI hasn't come for me yet.
george, my favorite elf 😂😂😂
1:14:28 “Dumb people talk about people, Mediocre people talk about ideas, and geniuses talk about next-level conspiracy theories.”
God has sent the manaul 1500 years ago. its still in same lanugage without any contradiction without any change of a single dot. stood test of time still relevant...all scientific text has been proven right by recent science. its called Quran.
Wonder how much Bitcoin or other mining added to the power usage
It didn't at all, rent free
@@macicoinc9363 not sure what you mean?
Human flops are being used sub-optimally. We have geniuses bagging groceries and idiots in big tech, "science", and medicine, because everything is about social connections, B.S. skills, inheritance, and lucky handouts.
But Ai Flops will also not be used towards some cohesive goal, they'll be redirecting in competing goals by humans with different goals. Ai Flops will be used by both DEA and Cartels. Porn and Religion, Scamming, and getting children more addicted to junk food. Ai Flops will be used by Ai to scam other Ai's out of money and computation. Power will centralize, then fracture when that centralized power becomes fragile and a clear target for other Ai's.
TLDR; Very Suboptimal use of Ai.
10:15 W is energy.time not energy/time
Educate yourself
@@DavenH I was high 😂. I'll keep it as a definition of "brain fart".
Ngl, watching a self-declared libertarian wrapping their mind around the very evident effects of late stage capitalism, repeatedly muttering "who did this", is endlessly entertaining.
Eat the rich 🍽️
What does late stage capitalism even mean? The government corrupting actual capitalism and turning it socialist and authoritarian?
Capitalism to me is basically just synonymous with consent. The government is abolishing more and more consent over time.
terry davis vibes
Intelligence isn't compression. Not well defined at all by any people in AI
Yes it is, to both claims.
found a video made specially for George th-cam.com/video/6-1oUMNX64Y/w-d-xo.html
i dont buy the 'aaa' video game argument. what if player 'god' slips up and doesnt play 100% perfectly, or what if its a 'black and white' type game and 'god' just gets tired of looking after humans. interesting argument, but i dont find it convincing at all. peace out. etcetera. ... ...
Lots of singing in this one…
A comment at 1:50:48 ... regarding web3 and tokens. You say anything that is web3 and has a token is immediately bad. Why do you say this? At the start of the cast you say the problem is people lie, the risk is centralization of trust. Web3 is the idea of decentralizing the structure, so no-one has root access to the server, database etc.. when done correctly and ideally this removes the need to trust people and instead trust algorithms and code to ensure data is consistent, not edited by someone without your permission etc.. So I am saying you are wrong to say web3 is bad.
Then when it comes to tokens.. this is just another mechanism to fund the decentralised network and provide the alternative to the old way of paying salaries to employees, instead rewarding measurable performance. The aim here is to replace the centralized platforms that control how shares are issued, traded and pay out and replace trust in humans with trust in code.
first
Instead of fixing twitter why won't you just help nostr not suck?
Pretty Gay
"it's gonna sky rocket" "but not fast" he speaks in oxymoronic paradoxes. not a great communicator sometimes.
is he a jew?
i like George but he's unimpressed conservative or close minded attitude about 9 things out of 10 is what he really lacks.
reason Elon is so successful is because he's good at seeing these kinds of things.
for example if we talked George about Electric cars before Tesla era he would be unimpressed and probably quickly came up with 1000 reasons why its stupid idea
GEORGE, what monitor do u use?
18:30 right is relative… jk