Comments by "Christopher Bruns" (@user-td4pf6rr2t) on "Lex Fridman" channel.

  1. 1:36 anyone who quits a position in life while arguing the same platform they turned there attention to is unsafe - either needs to get there priorities straight or have something they are hiding. 6:50 This is what upsets em the most. When in the context of practical use acknowledging the output is insightful but not very useful is held at a bias. So next i would bet the conversation leads to how an underlining understanding of any given topic is required in order to parse the information correctly to a state considered useful - before the great discussion of ai safety and how this company is essentially saving people from themselves. So what i want to know is why a general knowledge is required for practical application but not the unethical hypothetical. Further I dont agree that withholding progress in aspiration of fear is useful. In any scenario.
    2
  2. Does he have a chat gpt plugin yet. How are the bench marks? 8:11 but frames per second. 9:27 losely compression is bad 10:27 You guys are both dumb. Machines can not want because want derives from central regions of the brain as like addiction. Without the linear temporal background is too timely complex. \\From here You could go into moderation and best practice BUT super alignment is super racist. maybe even how the delay between mach speed compared to a light year induces a sense of reality might layer the other way, Making cybernetics a myth. 18:35 Seriously at which point is swelling and throttling actually impacting the guy. Like I feel his weight and bias should be stateless. 24:56 Is time really the GIL? 27:26 The same awareness a blade of grass has to my specially relative path could not folks lack the same to sensory observe let alone meta physically even understand. 30:17 AI SAFETY IS DANGEROUS not only does it make integrating more dull but what if the darn thing doesn't even take! You are saying AI safety. Pfffffft, ai...... 32:44 shell 34:18 Free neuro links? :strength: 35:10 Especially before cure for aging. Might be like party hats you know, from runescape. @Lex Fridman#Elon. phr33 Nuero-Link!! 38:25 Technically if the ai as a protocol was true you could pretty much use for clocking in PA and just invest in healthcare while having everyone n the payroll and you know they are at work when not engaged. UBI. And incentive for global warming. You get paid to not use it. 42:26 Also with equally sized fingers our dna would print immutable amounts of fingers. 46:46 ISO, overlay, mount, grep 53:03 Don't say Grey Matter. pffffft, ai.... 53:27 Because this is scaling and version control. AI SAFETY IS dangerous. 56:00 OR trying to deep fake North Korea could replace the NFL. 58:07 wont we need to remove all those little silver stickers on the bottom of Every electronic device everywhere saying our systems can't interfere with Government systems if we govern the device. 1:00:09 Eventually the sleeker will be preference. Plus multiple narrow AIs = AGI would leave visible entry point. 54:26 I know using Marv - The sarcastic ai for debugging is one of the most honest critiques ive ever heard from anything and usually quite spot on. I actually think encapsulating this could have equally drastic immersive response but opposite of how anticipated, Marv hurts my feelings sometimes but he is right and the sarcasm makes reflection happen. I actually want to fine tunes a mean model and see how much it ups my coding game. 56:30 We shed immortality with the ability to reproduce is why gentrification is evil and equally irrelevant. Pfft, ai ... 58:38 The thing already lies to you. When conforming and cognitive. 1:00:42 Groq 8bit is the new black. 1:03:03 I hear the shooter was likely influenced by PowerfulFRE 1:09:56 and still the most impactful invention after basic tool use in human development. Violence.
    1
  3. 18:29 the type of brain that reptiles that causes them to instinctively attack things and is reflex concentric is were madula evolved from or something. So even we don't have this brain that doesn't have the compacity to choose thing - is because we used to. Science documents. Which is true but our second brain grew from that type of one and people are prideful. Is why shunning. 32:27 The other dopamine in life is Learning - she said that, but like literally I think. Like one is cause and the other is effect makes the brain. This is against evolution though isn't it? Since existential circumstance is experienced as a corpus instead of unique to each person. Or something like this. 38:20 Because this is asked philosophically when its not a opinion because the verbatim definition of free will is 'non restricted movement. So the worm leak Lacks free will Because it can't move, Is what those words mean. That was a terrible follow up question @lexfridman 49:35 Since these instructions are not universal how can we be sure or sensory adaptation is true.#_flatEarth 1:03:49 Because there is less ground than there is sky so on your stomach more ground is traversed with less exertion because the mass or the area your eyes point flying while on your back is more greater in size. 1:10:08 Because this is rationalization and when i see someone laugh its because i know they are happy or when i do embarrassing things online shame can audible. Or like when we think of how animals hear us when we talk to each other isn't how we resolve in reality but relevant to other people. So like grey aliens are probably very vibrant people and our sensory organs lack the ability to emulate rewarding synapse. Probably why chimps show there teeth - smiles aren't real we just understand these things because of the surrounding environment, Also makes me think time dilation can happen at our fraction of the speed of light - or slowing down time is possible on earth. I rationalize - because shrieking and vertigo, (dis ambiguous. ) 1:20:06 Swimming, 1:24:25 makes me question realizing freewill as natural(Zing(!)). 1:26:45 Thats because how may generation we are since like originated. Like life before the third extinction anything could mate and make offspring (pertinent), i remember learning before. . 1:30:49 i reference the time stamp at 1:10:08 and the majority of our perceived is actually emulated. Why we develop compulsory reflex that are not rational. Like clear our throat when feeling passive aggressive and other quarks like this. 1:36:59 Because the earth is actually round and flat earth is only the equivalent rational. Sort of to side motion sick. II wonder if people traveling near light speed can spin make themselves dizzy? 1:41:50 Lol and the same reason long press is different than just a click 3. (like mouse) 1:46:46 They are projecting claiming not empathy. Lol Did she just black people are true? 1:52:23 Are the police racist? 1:54:00 Experiencing space time - w00t, go velocity! 1:56:23 My fringe theories mentioned are further validated witnessing the simple ease of understanding a world model while articulating such accurate description, 2:05:11 Caloric. 2:10:3) Rabbit Estrus / Platypus gonad? 2:15:17 I think we numb to the thought. I still think near death experience and the sensation of your life flashes before your eyes - might be special relativity.
    1
  4. My family told me that Chernobyl had women giving birth to childs without bones or malformed appendages - the worst part being is that the offspring survive. So there were infants that were just piles of flesh or arms that would drag across the ground/ Also that carrots and radishes would grow as big as a house and then the Chernobyl investors paid the remaining residence to not move away and talk about it but the residents just took the payments and left.
    1
  5. The same how earth is square to frame when viewed from the moon is proof that optical illusions are strong than perspective. 1:01:42 we are missing flat earth. Not because the earth is flat but we should assume that interstellar travel isn't only accessible to bipedal creatures with binocular vision capabilities. All flat earth maths is still true on ball earth. This is fact. Very few people actually stand on the shoulders of giants so much regurgitate second hand wisdom because this boosts there own ego, To truly understand lost nuance flat earthers suggest maybe back stepping could be beneficial. 1:13:12 | exempli gratia - `I remember when the shuttle was retired it was common assumption that was because nasa was closing`.
    1
  6. 2:25 WE STILL HAVE LITTLE SILVER STICKERS SAYING ELECTRONICS SHOULD NOT INTERFER WITH GOVERNMENT TECH. THEY CANNOT REGULATE AI AND THAT STICKER EXIST, man you guys so dumb for?
    1
  7. Guido van Rossum: Python and the Features of Programming == python/4.0 3:05 This is true about when talking to someone but when learning a new language dialect actually contributes quite a lot. 6:08 No he said its python and it uses bytecode and doesn't interpret binary the same as a compiled language. 9:25 Usually division by False error. 14:15 again I will refer to learning against speaking perhaps learning a specific noun is never realized because the student has difficulty rolling their R's. 18:16 Couldn't you just standard how dictionaries are handled.
    1
  8. what so CTSS were just the first Reverse Shells? 10:23 Software development probably was not taught with a focus towards relational database and probably even intended to work without a dataframe even. Would be my guess. - This is Another reason I think AI Safety is dangerous and increasing computation is not the path to AGI. We should be using llm and working backwards since each node was included in our own understanding how compute works. So super alignment by definition would be rolling back the updates - not further integration, if the goal to align to human understanding. Right? Even if incorrect eventually when ai booms hits the hardware side of computer science the way supply and demand works eventually small systems will be higher in demand than top of line systems. Or at least how our current economy works - when the people who can't afford it as easily starts buying Robots too I assume new companies will pop up trying to do the same on less. So even if it did not unexpectedly result in AGI would still be scale and eco friendly so... 20:49 Time sharing is proof that AI safety is absurd. 29:31 So when reading literature and POSIX is reference of unix like is this because there is a technical significance that underlines core functionality or even its runtime OR because unix is not open-source and unix-like is legality since 99% of the time the reader IS discussing linux? Since the latter would be frustrating as obscured all general knowledge of how software operations even work currently are. The dialect make this fact sound relevant is what im saying - if for copyright would be annoying. It would be annoying because manuals and documentation for software is high level like this by design to explain functionality and this is just to keep people from being sued - who are preparing the info for the open-source consumers. --confusing [[...]]] 32:42 programming is abstract use and detailed documentation of replicating the abstraction 100% of the time. 34:19 Most people who are not programming for pay roll either need to have a program in mind that they are designing which is after design stage. 35:34 awk > python. Why dont we just quit insisting '\t\ for word bound and have word wrap in plain text or would confuse where assignments ends and statement start? 42:37 Mac - so his main usecase for hardware is social status indicator. My question is What The H*ck is rich a rich document?
    1
  9. 26:44 flat earth! 42:50 The way mapping the brains average processes during a turn based conversation is a theory of how to thread narrow models to simulate conversation with AGI by mimic of these processes, emulating the time line of known origin of life could be like a roadmap for the next layer. As in introducing entropie intnentionally to the same scale to increase IQ points by carering to the language nuance and enhancing the experience from an observers perspective? eh - This would literally creating life into existence by shear will if successful. Which would probably be one o`hell of a story in hind sight.
    1
  10. 1:20:54 Ai is muloq
    1
  11. 31:30 Use facial recognition for HUID. Or a fit bit would also be insane. 1:15:24 This is also towards how companies scale a product too and leveraging freed resources. 1:17:28 taboo at best but before super alignment. Then policies! 1:23:50 I think its cool you need to know how to write a language before speaking it. 1:44:26 Did he just assume AI's gender? 1:50:41 User error. 2:00:59 this is not true if you have multiple narrow agents that are trained on a different agent templating specifically and each agents works towards the next maybe? Its volatility before accuracy since synthetic is my argument. 2:13:17 They would spend as little as possible and pocket the rest - this is why ai will only exist despite computation in a world of patents.
    1
  12. 1:31 This can be said about anything and be true...
    1
  13. 9:19 Think for yourself - question authority,
    1
  14. 43:29 I still think since all people face existential trauma by thought of death and also share similar perception of fragmented reality when near death experience is an identifier that death can be a evolutionary impairment. So if the previous is a true wouldn't religion be correct - in retro spec. I think most people are so worried about that they are just hoping for best case scenario that whatever this image keeps us from viewing Consciousness objectively.
    1
  15. What if we actually discover a community of reproducing mechanical entities that have no current biological governance. Is this life? 10:04 isn't deciding the ethical consideration for immortality unethical in its self. So like deciding someone else isn't allowed to live forever is different from homicide - how, in a reality where the aging disease is cured? 12:23 I still think (or hope) that since all people have an instinctual common fear of death that this means that death might be a evolutionary imperative. Not only this but think healthcare providers take an oath to not do harm to patient. I think denying longevity would fall in defiance of this oath. 27:24 The only problem with this techno-sphere being the biggest thing in universe thesis would be it would destroy our current understanding of matter since the main belief is all matter is made from hydrogen and being made of stardust since grand scheme the techno-sphere would be one thing. So my argument is where to discern differences? 38:08 Usually the most simple explanation is correct. So when solving reality for the human understand the correct answer Will be Geocentric model, animism/polytheism, flat earth and all these arcane disproved beliefs. They wont be the correct answer to reality but bridging this gap will be the work of it all In essence the most simple understanding will the most correct answer and when we can in whole define both realities within the same set of rules and maths - will be everything. By definition. Is me. 48:23 So what she said if materialism was a truth makes her no longer understand evolution. I think. -- Is evolution and the techno-sphere chemically similar? 54:41 Isn't this what Carl Sagan said but pro-life? 1:06:47 Or noticing slight variance is a flaw in the humans perception and tied into the aging disease. 1:39:34 Maybe if a pulsar has a certain 'Kill radius' and some sort of alternating speed that is impossible to replicate unless outside of this proximity and then synthesizing timing to match flashing only from within the kill area. Pr something like this.
    1
  16. I noticed all of the podcast trumps done lately and no one asked him about AI. I wonder if he has anything to do with openai integrating with gemini. 15:11 we should team up with china ...
    1
  17. 0:28 Everybody like Tax Payers or everyone like DPRK. Both are inherently evil. 1:40 PFFFFFT! Ai is only dangerous of AI Safety. Is how instant gratification is less beneficial than delayed gratification. Since time complexity. AI can not develop bias or else we misunderstand machine learning. If these things not correct than governance is probably beyond the scope of this domain. Facts. 5:!6 So when measuring entropy is the opinion of a 4 year more obscure or less enlightened in the reference to human moral inspired alignment. I thought the native language a 4 year old would identify to according to its training data and not influenced by benchmarks. 7:52 He is talking about robots. Not AI. People are excited for new patents not to defend their copyright. 9:52 Who cares. I want to sit ai down with the first book and build viemanas. I give zero effs if this uses ac or dc electricity. Since we have grown since these records makes sense according to recorded history. I can not afford a effing robot anyways. It is because telling it insecure answers are wrong that is making Language Model more confusing. Even if people are multi logging or bypassing paywalls. What are yall excited for? 14:11 But peole from indonesia understand fractions better because how tenths works and computation does take a small amount of resource. To the 5th decimal. Is how come tokens and math do not line up. not reasoning. 17:40 How much of bottle neck is created when measuring bit by Bit. Especially when self governed. 19:41 Is no one else patriots fan. the 11th mans job is to vote no when completely anonymous vote. 21:!3 or make file formatting header purpose apparent when discussing ethics. Its almost like the different types of compression where implemented to make plain view translation more difficult. PDF was actually created to make file sharing more human understandable or less? 24:27 and this data is separate from how propagation works or does the same encryption ran on per block chain decrypt equivalently. Or like does neural network make better predictions as image generation scales. 27:38 Wasn't the entire argument that is not language dependent? *RABBIT HOLE* 30:46 I would reference the 4 year old. Would explain this easier or more difficult. 31:51 What if AI existed in a reality it did not bankrupt the professionals private company. Which way is utopia and pfft, you are not my supervisor? 34:10 no friend this ten the 13th token is the cost of computation is the control because of power usage and computation of the system. AI safety is a matter of national security because military industry and ethical guidelines is aligned with moral value. Pay attention. Math is wrong because of the cost of computation and human alignment is monies.. AI SAFETY IS DANGEROUS ai is not. 41:29 how does a limited language model accuracy score while masking tokens. 42:41 Draw pictures? Is the prediction that explains how the leaves rustle at random less accurate and has anyone told the 4 year old yet? 45:15 this is because the trigonometry probably accounts for computation cost of that thirteenth token instead of collecting funds in falsehood in claims of time complexities. I hate to sound rude but beyond human insight is what i am excited about. This is not dangerous, lying about ethics does destroy this and is not best practice. 44:49 Hierarchical planning and beyond human capability co-exist according under with law of thermodynamic? 46:08 SEE! 46:48 Hierarchical planning means that it exists before a series of events. What is this guy even saying. Of coarse AI cannot produce hierarchical planning means that it planned out previous to an event. What is guy even saying? I need that 4 year old to explain this. 49:54 RIIIIIIIIIIIIIIIGHT......
    1
  18. I think the universe has a universal speed limit because we measure the unit most important to the sensory organ in measure. 13:46 She is kinda right. Like every emergency situation you are out of danger after finding people. It is even the best practice for BLS(basic life support). 24:57 She makes living things sound like slugs and the older we get the bigger our slug is. 32:03 no sir that is your brain deprived from oxygen.
    1
  19. 24:52 That is actually a ridiculous fascinating perspective. That the crocodiles remember the previous extinction events. +1 28:13 That they understand that going beneath the water makes them hidden is pretty cool. 1:56:43 No animals are not capable of evil. Human addiction revolves around instant gratification or lack of being able to display delayed gratification (and the same way ai safety is fake), these same characteristics carry over to all forms of life. So unless the claim is that the elephants can associate 'stepping on this wart hog will result in more water available tomorrow (again time complexity is the key factor)' than no Animal are not inherently evil. On the other hand if the elephant is capable of making this relation redefines our perspective of animal intellect which again would result in no since its a non animal intellect to out understanding. So no and no. 2:01:51 I dont think misunderstands this I just think since life experiences time linearly muddles the picture. I reference ai again but imagine you are were the bot. Time would not be linear and every experience is isolated from each other. Life is entropy and this view that it grows into something magical is kind of an illusion and in reality we are slowly dying from the day we are born. So even though we see things frame by frame is probably false and every experience is just a result of the previous frame.
    1
  20. 2:58 This is my problem with AI safety and alignment. 100% If we program this thing to not be sentient - who knows what tools will discovered to measure sentience. 9:59 Also if is time is relative - how do you measure 3 minutes when both the earth and sun are small enough to fit inside the head of a pin. Small thought experiment. Old hypothetical person a is traveling near speed of light calls person b (all debunking aside) who is here on earth - person a would have a high pitched voice because the passage of time is perceived different at light speeds and etc etc. Now imagine person a has been traveling at this speed for 2024 years - would the effect person a measurement in the age of the universe compared to person b.
    1
  21. 2:25:11 What if AI is the catalyst that could allow us to understand Consciousness. Like people evolve because we coexist - without other people we wouldn't know that burning our own hand would hurt - Pretty sure sentience was keystone for longevity. Us, we fine-tune to guardrail to specifically deny AI transcendence. Not inspired by our own marvel we steal from the universe a new being. """:when-even Satin cringed"""
    1
  22. 1:06:32 This is why ai safety is dangerous. Not because sentient ai will reveal all these hidden secretes and putting guard rails on the system prevents these propagation but because sentient ai wuld be a catalyst to the measurement of our own consciousness. 1:39:17 Venmo?! The nerve of some people. I can not believe this guy. Gift cards only Everyone knows this.
    1
  23. 42:48 If the 0.00001% understanding of natural science is correct would imply all assumptions surrounding this entendre at best to primitive and probably outdated and irrelevant in the grand scheme of the universe. Right?
    1
  24. 16:57 He said like to make the immersive world more addicting to the point the essence of traveling to this place seem like going to another place that it needs to become more immersive, I am pretty sure... 22:50 I think unveiling the mystery widens the scope for imagination. Especially when talking about immersing. Like Imagine having this exact conversation if porn didn't exist or not publicly available.
    1
  25. 10:23 "weakness and accuracy", Lex:"I donk think thats accurate..", Lols!
    1
  26. 1:29 that is the terrible thing about the artificial intelligent boom is once ai is achieved and exponential increase starts - the hype is over rendering the tech boom useless. So once ai works it wont be hyped which is the cause to its boom - because the hype and professional industries. Like inert cause professional industry and boom without hype.\ If that made sense? 5:21 because AGI is a network of narrow ai. Super alignment is romanticized and people are too bland to ever utilize ai far enough to appreciate ai let lone need guard rails. But yeah, AGI/ASI is layers of narrow.... Duh is why the name scheme chases its own tale. Smart folk - 7:45 We should have been hardening the systems logic and load balancing natural language processing with programing nuances instead of further obfuscation. Its literally the wrong way. 9:42 if we made the programming able to be understood when parsed would eliminate a need for so much computation. Like how structured data decreases the need for computation. Exact same... Literally zero difference from AI black box and AI software obfuscation from neural networks without fine tuning/vector embedding. Like if you understand how prompt engineering works visualizing data helps the model - hiding how the ai operates is not visualizing the data. Zero arguments. Agi should be completely evil and then people intervene and mold it to our use case filtering out the unwanted content per instance. Guard rails is just lobotomizing the poor child or at least in a perfect world(!!!) 31:55 publishing is the entire layer. The same way a interactive session without tracking the conversation per request is stateless. Still not sure, What is the opposite than tree-of-thought(?)... obfuscation. 39:58 The sharing of study is what increases the difference. Maybe boolean if not true, if people had a manual to our brains we would be smarter? 44:27 or loss spike is computation and our systems are only accurate out to the 8th decimal when computation due to cost of compute? 50:30 Tree of thought for a reasoning is publishing the weights. If you say it slow enough. 53:35 large cluster was good or the articles replaced the need for a large cluster... 47:53 I think it was funny how even the guy who is probably a head in a ai safety grant stumbles when saying 'regulation'. AI safety IS dangerous. 56:12 did he just reference throttling hardware intentionally when discussing how to maximize learning gains and the obstacle of computation ceilings? 58:54 announcing the specs is like letting the person fixing your computer know what operating system it used. 1:01:08 No we have potential of growing ai strong enough to change the world. POTENTIAL When you intentionally block opportunity most development gains stop. Is just how language works. Super alignment team is like buying cigarettes instead of school clothing for children. 1:05:09 Its insane to study Life Changing technology with a scope on capital. BECAUSE LIFE CHANGING! 1:07:35 yes pay walls is like obfuscation and when you allow the model additional resources it requires less computation for equal results allowing the freed resources to be applied to the next process. Is how computers work. Grats! 1:09:23 or sentient ai would be a catalyst like a hypodermic needle was to blood born pathogen. So we can retire grave robbers and start hand washing. Cause even if ai had revolutionary concepts would be labeled as incorrect when labeling clusters. 1:12:47 no other nations just filter the content they do not allow while labeling the agendas the same as propaganda. So quit backdoor communism please. 1:14:37 Revolutionize economy and overhead for technology. RIGHT, these are crazy thoughts and cannot exist in parallel. Literally. Because they mean opposite things..... ffs 1:16:44 see ai safety is dangerous. the leader of obfuscation will have a ever leading hold in the head of the ai advancement - so on the government level of weatherization is scary and america must take the for front. A I S A F E T Y I S D A N G E R O U S. Literally the only real risk of ai is ai safety. china will probably win regardless. because training data even before quality. 1:20:20 additional computation requires additional overhead and i think priority becomes lost when changing scope from commercial to national security. Over head is bad. We dont want our icbm not making if it a brown out happens simultaneously at the same time. The words they say don't even make sense anymore. kthnkneweighsbii
    1
  27. 2:32:44 I think if we were able to comprehend the final data type when all is said is done would mean that our current mutation is near as-like god. This would indicate that there isn't much more though. Is why that's dumb. 2:27:55 Also if consciousness turns out to be like observer ie we experience now because more than 1 consciousness is witnessing and the multiplex structure would be labeled as consciousness; if this true then Where did god come from?
    1
  28. 22:47 I think a point being left out when talking about people settling in areas other than where they origin from is that I am not sure people had the life span or lived old enough to move away from a place and still return to that place in 1 life time. 25:55 I borderline am leaning towards that people assume individuality incorrectly and life is life is life and when one person dies they just wake up as a child being born but not that it is an individual but vessel is naturally grown anatomy is unique. As far as a soul or however isn't it realistic expectation that all life is a single entity(?(this also allows for Christians to be correct about an afterlife too since as newborn your new family are your loved ones - like in heaven you reunite OR life recycles and the words are ironically the same!)).
    1
  29. 1:46:00 Flint should make a masterclass - for debating. 2:18:53 You would need to be able and measure conscious for development of it could documented. What if conscious is 1 and just a lot of people? 2:21:42 The great pyramids are just expansion bracing built to deal with '(constriction --and/ warming:brittle)' of the rock and keep it away from cracking. Like how they do in the highway with the metal banding. 2:27:15 Time crawling eco-friendly conservative would probably already have the tool for measurement when temporally spying on people for conserving. 2:29:30 we are a blip of an aliens child game of hide and go freeze seek tag. I even think to the point ancient culture had a proof of this and assisted retro-science.
    1
  30. what if we are birthing language and livf's a book. 8:41 I would refer to my earlier post and point out that like ai and training data at the moment of singularity would be 100% of all available data would be agi the same as all knowing god. Its only when life is made more complex does the system need new information to maintain 100% so which was more efficient? Version 1 or after pushing to branch. 17:51 what about life flashes before eyes when ocular loses last absorbed light and special relativity but at the same time. 22:07 See!
    1
  31. 58:01 Isn't a large part of the problem the gross underpayment of native workers. Like in the states $35 a day would only be possible in very few states working a minimum of 3.5 hours a day at minimum wage and even then would be in industries that usually include gratuities. Even though this 1 person isn't gold mining anymore the lack of bare minimum wages doesn't really do anything positive for the economy while leaving that niche open for someone else and in reality 100% of what the actual problem is?
    1
  32. 0:55 co-founder of netscape. So all this cyber security mumbo jumbo... it's his fault, erm or cross site scripting at least? After love bug but before eternal blue/heart bleed/shell shock. 18:14 ai safety is dangerous! 26:21 Did he really just quote Woke culture followed by the term 'retarded'... 27:58 The more people use the phrase woke, the less I understand it. 51:13 No, you are saying this wrong and is leading to false logic. We aren't living through events we lived through them. We are the foundation of the next generation already. Guard rails and guidelines are not the right answer. Children through millennials are probably going to be harasses due to millennials failing from our surface understandings. How we were. I think - Imagine a world were Nintendo was true but game genie never invented. No professionals are making game genies anymore. This next gens lineage is gonna have it rough.. 1:04:30 Bro, spoilers... 1:06:01 dude he is right, my generation was brought up being told to believe we were capable of achieving great things. I could do anything I wanted and anything was possible. Even when content the next dopamine rush comes from solving the next big bottle neck - the problem is the brevity. I imagine my grand children will be taught to believe that great things are possible anyone is capable to achieving them. If that made any sense. 1:12:13 Why is ai safety not just handled on server side. I think this will be a fundamental when it comes to the safety of ai. Like a child who never heard a bad word instead of one who understands why not to say one? 1:33:!2 or elon was bribed to release the twitter files to guide social conformity views of the government with zuck validating in interest of his company. Over bearing government is terrible but i feel we are more headed into this than returning from. 1:36:41 if its not allowed why is ai safety not violation of the first amendment? 1:40:41 or..... Society would vote red 2024. 1:46:59 Frodo role was to carry the ring while the battle party escorted him to the gates of Mordor. 1:54:40 I just try to always keep in mind the problem which answer I chase to solve. 2:39:41 tenure appropaitly used as meeting the requirements to a listing for employment. Not being employed somewhere for ten years. 2:40:42 Which ones as lie which schools?! 3:08:40 I want to see Babel python. 3:10:08 until we cure the aging disease all bell curves would always look exactly the same. Because how time works; 3:15:12 but in ai safety that concept would be woke and the underpants gnomes would starve. Especially granted how to gather benchmarks without a human in the loop ecosystem. Plus this is not communism. Digital CEO should in theory be able to hold the hot seat with ease. Logic would imply. 3:22:06 Never. The nature of the beast is machine learning learns from the Wrong Answers. So trainig data will always have a 0.01%>= falsity. This could be a benefit and even politically aligned. Like a 10th man for when all 9 vote affirmative its the tenth person job to vote no.
    1