I think increasing your iq to 190 is infinitely more triumphing than becoming rich. Why not do something everyone thought was impossible? Why not be the first to achieve something so great it’s revolutionary? Why focus on silly money that everyone does everyday. Become the exception!!!!! Become remembered and known through history! Give the world hope in becoming smarter! We are rooting for you!
I study software engineering, have created machine learning models, have been to conferences discussing the topics and I have talked with experts on the topic. No, AI will not replace all jobs in the next 5 years, especially jobs requiring high IQs. AI knows a lot but is not intelligent. AI is kinda dogshit compared to us when you think about it. ChatGpt needed 30M$ of electricity to get trained, I even think thats a lowball, and cant solve a lot of logic problems that most humans could solve. AI consumes staggering amounts of electricity and requires a lot of hardware only to be worse than a bee at image detection. Thats the situation as of now and if you dig deep, as maikeru01 said, its just numbers it only gets fed numbers and outputs a predictable outcome from those numbers, theres no logic there. It knows a lot but cant innovate, it cant invent new stuff because that new stuff would need to be fed into him in the first place for it to know about it. Even if you dont believe me on that, just imagine it does start to steal our intellectual power. Dont you think governments would step in? Dont you think we can just cut the alimentation from those AIs and thats it, it cant do shit anymore? There would need to be major fuck ups and loss of our inate need to defend our natural intelligence for AI to take over the world and control humans as some of you guys said.
As i said in the video, AI does not need to be in control for this to play out, super intelligent AI controlled by humans would result in the same outcome
@@InfiniteIQ Yeah and I mention that governments would crack down massively ont this if it was the case. Even if this doesnt happen and companies are allowed to lay off all their workers and everyone is now replaced by AI, companies will still need to sell their products, the products will still be there, the economy will adjust and if we produce so much with ai with little human involvment we will be in an economy where people barely need to work to survive, wich is also a good thing. Its a trchnological revolution just like the internet wich automated a lot of things, markets will adjust.
@@InfiniteIQ I'm studying computer science and have my undergraduate project based on AI and it's really underwhelming how simple the task is compared to how complex building it is. But I agree, it's like the invention of the gun is to war, as AI is to business. If you can get more information, and transform that into knowledge in the business, the first person with the gun will have a gigantic advantage like Americans did in early history. They will most likely get a boom. But to be fair as soon as businesses get AI to be at this point, competitors will follow but be at a major disadvantage. This is basically data science.
@@InfiniteIQ I've had similar concerns in the past. My IQ was naturally 157 before I had a brain injury, and I'm about one and a half years off from recovering from that. I've not yet taken stock of exactly what I've lost, or where I am not, as I want to 'feel' whole before I measure where I'm at again yet I digress -- AI should shrink money making opportunities for people of this level, but it's not going to be apocalyptic. AI in this sense fundamentally becomes a tool. You mention in your videos that having high IQ has given you an abundance of good ideas, and "High-definition vision." I had both of these things before my injury -- the former has somewhat returned, but the latter is pretty much gone. Being where I am now, remembering how i was at my lowest, and at my peak before all of this, I can confidently say that anyone in my former shoes, the 157 IQ ones, would be able to use the 'tool' of AI, or even just formulate their own organic ideas into concrete plans to execute far better than someone of a lower IQ using AI as a tool. AI can expand the breadth of what someone like this can achieve, but depth is a hard ceiling -- AI can feed someone as much brilliance as it can, but if they cannot cognitively digest it, any product they output is going to be at best, inefficient, at worst, totally broken. Check out a lot of the AI trash on youtube right now for example, but imagine that extrapolated to every nook and cranny in a modern economy. The bigger concern is rather governments shutting out new opportunities but whatever -- point is, have faith in yourself. If I could be how I was again, I would not have a worry in the world. Everything was easy, and I felt comfortably isolated at the top of a pyramid.
AI is a scam. It has no intelligence. It is difficult for me to articulate what AI is but I'm very sure about this. When AI generates something, it does not have any understanding of what it is doing. For years it was outputting random images and humans would tell it which one was better. It shows you one cat with 4 legs and one with 6 legs and you choose the one with 4. And it removes the idea of cats having 6 legs from it's data base. But really it doesn't know the car has 4 legs. To a computer program an image is just a series of pixels. Not even pixels but code for pixels. And it has lists of patterns of pixels that humans have said are good and lists of patterns of pixels that humans have said are good. Same for audio, same for text. When you ask chat gpt a question or doesn't understand your question. It is replying with something generated to convince you that it was written by a human. It was designed to pass the turing test not to give accurate information. So you can ask it to give you code for something and it has a data base where it sees that code looks like this. Code starts with a left carrot and then it has these patterns of sentences and these file extensions and it's formatted like this. It will give you something that looks like code but probably doesn't work. You can ask it all kinds of things and unless it's really simple it can't give you any good answers because it isn't thinking. All it's doing is scouring it's server for replies that look like they are supposed to follow your message. That isn't intelligence. It does not have the potential to improve. It will not ever do anything beyond this. These companies are scams to trick dumb investors into giving their money to Nvidia, Google, Twitter, openai, etc. if you have stock you should probably dump it before the crash.
That's AI rn most people saying that kind of stuff fail to understand that your brain is basically the same, but infinitely more complex. as the AI shit gets more complex (A LOT MORE THAN WHAT WE HAVE RN, LIKE AN INSANE LEVEL MORE), these sorts of arguments will lose ground. of course, that might happen in 2 years or 200 years. who knows
AI is just probability where every node is a floating point number, so instead of definite output (This is a dog pic) it will see input of a different dog pic and approximate the result. So you can have a different angle or breed of dog and the AI will still know if its a dog. So a dog from a slightly different angle is less than 70% a "dog" It's never fully sure or not fully sure it's just a rough guess, from 0% to 100% certainty It's not really a thinking person, it just takes in a noise texture (random pixels) and moves the pixels into the right place to roughly approximate a dog (probability) It does the same thing for chat. A "dog" is just a large database of dog pics that are recorded as numbers. if you get a paper and drew some lines, the AI would look at the lines and see if it is at least 20% of the lines that is supposed to be shaped like the dogs in the database. It will think "whats the likely output for" h e l l o. Otherwise it will say the same thing over and over like it does. All the AI sees is random letters (Goodbye) and looks for the likely letters to respond. The AI may give you the wrong answers like fake names and places, because it's supposed to produce the most likely answer for the letters you provided. like writing the number (6) that not pixel accurate it will still know but if you write an (s) it will know it's not a (6) ============== With character chats, the AI has info of words and sentences that personality said in it's training data. And it has respond with letters that are a rough approximate of the words used in it's training data for "aggressive" Like saying "Hello! i am the banana man!"
@@LOLSflint i think the difference is a human can run into a new situation that is similar but not the same as a past situation and come up with an idea for it. or a human can say something that they never heard someone else say before. the ai actually can't. if you ask it how many spark plugs a Lexus ES350 has, it will look through its database. if somewhere in thet data base it has an instance of someone asking this questions before, like maybe someone asked on reddit and a few people answered and said 6, then it can reply that it has 6. but if you ask it something that it doesn't have the info for, like if the car has any PFAS material in it, it will either say it doesn't know or it will give you an answer that it thinks sounds like a human answer. It may very confidently tell you that the car does not have any PFAS materials because it saw other instances of people asking that question about other products, but there is no link between whether the car has it or not and what the AI tells you. And there is no way for the AI to find that information, and the AI is not programmed and does not have the ability to simply connect to the internet and look for the owners manual or something like that. Whatever data is on it's server is what it has access to. It doesn't know if it's lying or not. It doesn't know if it's right or wrong. It's just generating the best reply it can. It will not be able to find patterns in sports betting or day trading. And it will not improve it's own software because that isn't what it does and that isn't what it can do.
I've accepted my future as a serf under capitalism. I'm just gonna try to spread good karma in hopes that we will eventually reach an enlightenment after the techno-feudalist dark ages.
Also please share your thoughts on Lion’s mane mushrooms there are many videos saying that it will increase your brain neurons. I figured that would help with increasing iq since memory is improved. Perhaps cover it in a futur iq diet video?
@@InfiniteIQI watched your video and you said you’d be interested in taking it again for a longer period of time to see if your body would adjust to the higher number of neurons that you hypothesized. Have you gotten back to it?
@@4unoahthere could be rewiring and stuff, but no new neurons are formed in humans and apes after birth, except in the hippocampus which is long term memory and doesn't have logic gates. I haven't watched his video but I'd confidently bet money you misunderstood him.
Requisite variety and kolmogorov or descriptive programm complexity are key to understanding why we won't be able to achieve technological singularity. The system/programm cannot increase its complexity on its own. Complexity is the closest notion to intelligence as we understand it. Thus, we can likely not increase the intelligence of machine in an automated way. We can make them more and more complex but we cannot build programms that increase their own complexity. It is incomputable.
OK, but then you disagree with general statements made by Stephen wolfram who analysed these networks thoroughly. The conclusion is overwhelmingly speaking against the possibility of neural networks in the current form to handle irreducible problems. Neuro symbolic computation may resolve it to some extend. The core idea still remains. These devices cannot possibly surpass the laws imposed by Turing machines as they are build on top of them. Problem number 2: we can easily state in fuzzy everyday language things such as efficiency of neural networks without being able to provide a correspondence/isomorphism to respective low level Turing machines. Neural computing is still based on the classical paradigm. They can't surpass the computational limits imposed by the halting problem. You would have to give a formal explanation what this efficiency does in the context of computability? Is neural efficiency not predicated by the amount of neurons? Would more neurons not correspond to higher variety and its encoding require higher descriptive complexity. Could you build an efficient neural connection based on just a few neurons or a thousand? As far as I remember, more than a billion weights, respectively connection's were required for chatgpt's functionality. Pretty high variety if you ask me! The law of requisite variety and its related notion of incomputality just make the claim that once you designed the algorithm and run it you can't make it more complex. In order to do so you would have to jncrease its number of neurons possibly needing more storage. This process can be infinitely extended by humans. Once you start the device however at the level it is , it can't become more complex. More specialised and refined or adapted through training. That is possible. However, not more complex. You would have to provide a formal proof of things being different. Cybernetics and theory of computation are the only valid theories so far we have for describing computational processes. Everything else are speculations enabled by the fuzzy nature of natural language. More than 2000 years of lack of results in philosophy showcase this. We can't go very far with ordinary language. The major reason why chatgpt seems so fantastic. Let's phrase it differently. It's easy to talk the talk and difficult to walk the walk. -'The map is not the territory'
I get the idea that you are referring here to speed. TRUE THAT. However, without sufficient variety or complexity you cantvdo much to begin with. A simple neural network may be very efficient if run on a photonic computer but it can only perform the operation 1 + 1 quasi infinitely faster.
@@InfiniteIQ Efficiency is a key attribute of intelligence, involving performing tasks faster or with minimal computation. Irreducible systems (e.g., chaotic systems, P-vs-NP conjecture) pose limits to computational reductions. Computational irreducibility suggests some tasks cannot be simplified. Efficiency is more applicable in reducible/regular tasks and complex tasks with regularities. Building efficient neural networks can reduce neuron usage, but cannot go below the minimum variety needed to compute a problem. Using a single neuron for efficient computation is limited in certain contexts. While efficiency fits within IQ research, it’s insufficient for rigorous computational definitions of intelligence. Precise definitions of intelligence remain elusive; current notions are ambiguous. Ross Ashby’s definition of intelligence highlights the balance between efficiency and variety in neural networks. Specialization for efficiency can lead to a loss of adaptability in systems.
@InfiniteIQ I agree that AI will largely replace humans in many areas. We can automate most, if not all, things we have invented and formalized properly. However, we cannot automate the process of automation itself, as discussed earlier. Over time, AI will likely replace a significant number of people. Remember, AI is the world's best copycat-far superior to us. Even if humans come up with new practices, they will need to improve efficiency, and AI will likely perform that task better. This doesn't mean humans have no place, but it does mean that most of us will be replaced unless we pursue becoming renaissance individuals, which most won’t. It also raises questions about the kind of world that will emerge once AI is fully integrated. If wealth is not distributed adequately, we could again face a scenario where a small group holds much, and the majority has little. The idea of AI replacing humans due to efficiency gains is widely accepted, but it’s a short-sighted view when considering long-term economic stability. Unfortunately, as beings who learn through trial and error, we may drift in that direction.
Hey man, I know your goal has changed but it'd be cool to know when you're planning to take the IQ test again. Probably after quad 9 back after your 2 year money making period. If you're planning to not take it too many times, that's also fair. If it were me and I was doing an IQ raising project I'd like to measure that as frequently as possible without ruining the results.
IQ tests can only accurately measure up to 160 so there is little point testing when i am already 150+ but i will test again when i hit quad 9 back to see if i am at 160 at that time
@@InfiniteIQ There are some IQ tests capable of measuring above that. Try doing the sigma test light by Sigma Society. Its an ultra high IQ community created by a brazilian genius with an IQ of 233. And please, reply to some of my comments towards you in other videos, there is information that I believe you will find interesting.
Ai wont replace construction any time soon. I think it takes more time than five years for AI to implement in our economy. And you can do thousands of Iq tests and improve your iq. But that doesn't mean per se you become more intelligent. Because if you study iq tests, you mostly learn the similarities among different iq tests.
@@RedditTales_009 you cannot. I would like to understand what you even mean by IQ? Could you exponentially increase your knowledge and skills through memorization in specific fields? Yes of course, through different methodologies you could increase your knowledge and also expand your perspective and way of thinking, but if you by IQ mean actual processing intelligence that is mostly genetic, meaning it depends upon your genetic material, some people will always be inherently more intelligent than you, they surely will not be as knowledgeable but more intelligent for sure.
I agree with you, we have still not realized how soon this change is coming. AI's invention is akin to the discovery of fire, drastic changes will follow suit and we need to prepare ourselves for that.
"Prepare" in what way and to what degree can you prepare for such an impactful hypothetical future AI will have? Any degree of preparedness will be barely relevant compared to someone entirely unprepared when it comes to being effected by the future aspects of AI. If AI brings upon a golden age of technological and scientific advancement solving global crisis and achieving abundance of resources we can expect the level of lifestyle to grow in quality, if it takes the polar opposite direction we could live in a super intelligence police state in which we will always be outsmarted by a 10,000 IQ super intelligent entity. All we can do is see it unfold, the general population even if aware would not be able to do much to prepare.
Thank you,Daniel.Please also make a video about best skills to learn during AI era? I mean,certain people also will get rich during AI era and it is necessary to learn skills that will be useful during that time.Please make a video about it that we won't miss the train when it comes
AI will not reach super intelligence, or even general intelligence, anytime soon. I almost want to go as far as to say it's even an impossibility to achieve either of those things period and that we have made a categorical error.
Focusing on making more money is great, but it would be even better if you created a second channel dedicated to business insights and strategies-maybe call it 'Infinite Money.' However, since this channel is named 'Infinite IQ,' we expect content that delivers on that promise. We'd love to hear more about techniques and methods to boost IQ, and we would deeply appreciate regular, actionable tips on enhancing cognitive abilities. Keep the content aligned with the name, and continue sharing new ways we can improve our intelligence.
I am not here to serve anyone, i am not interested in growing on TH-cam, it is worthless to me. I just enjoy sharing my ideas and what i am working on and learning, and if people get value from that then that is a bonus
Because ai will as you said be controlled by people and will not be able to develop true agency on its own there will still be lots of economic inefficiencies. What will occur is actually the average man will be able to access ai more in the future, and will be able to develop businesses a lot faster. Intelligence will be the factor for developing the ideas which you then have an ai quickly do the technical work for. We already see this occurring with complex models being developed by chat gpt in firms which might have taken them months to formalize on their own. It can create a golden age of opportunity for the middle class and arguably already has.
@@InfiniteIQ supercomputers will have competition from geopolitical competitors and moore's law will over time make the tech viable for civilians to use, similar to how 1960s supercomputers are less powerful than your smartphone
Companies relay on employees to get paid if all companies have no coustmers(employees) then who will pay them if all companies use AI Thus Ai will not replace humans unless its self driven then that will ofc be proplem
Companies relay on employees to get paid if all companies have no coustmers then who will pay them Thus Ai will not replace humans unless its self driven then that will ofc be proplem
@@InfiniteIQ I don’t think so i think it will happen but it will make the system collapse because companies relying on employees to gain sales and employees relay on companies to gain salaries if any of both sides fall both do so
Если ИИ будет настолько хорош, ему не составит труда уравнять людей которые разбогатеют к тому моменту в пользу своих разработчиков и правительства. Надеяться что мы будем играть по тем же правилам в эпоху ИИ что и сейчас это наивно на мой взгляд. Это просто будет новая реальность как после падения астероида.
Do u do other brain training other than quad n back? I mean quad n back trains certain types of intelligence (namely/mainly working memory capacity) but there is more aspects
What if we were to create paper markets? Like literally, go back to basics to stop ai from taking advantage of the algorithm on a computer? Would that not work?
What should I get into then? I’ve been thinking about learning trading and algo-trading after that, which AI will surely make much more difficult than it already is, but I know some algo traders and it can be highly profitable once (& if) you find an edge which takes a couple of years usually. The sports betting thing seems interesting, but it seems to me like its just gambling. I’m 16 and broke so no money to start with and parents didn’t allow me to get a summer job My goal is 10k/month before 25
@@InfiniteIQ Daniel, I hope you are doing well. I have a request: can you elaborate on the words you spoke in the previous video about achieving means of living off of passive income in 1.5/2 years? I wish you would make a video about how you achieved that. I greatly appreciate your channel since our goals are identical. Please, excuse my errors of grammar.
Well, with super intelligent AI there could be many scenarios. Lets take the hypothetical scenario where AI becomes smarter than a human is controlled by a human, and serves as a very powerful money making tool. Even though its less likely that something actually smarter than a human would want to remain subsurvient to humans, but lets assume its possible. Why would that make you change your number one priority of becoming more intelligent into a goal of making more money? Why would a super intelligent person like you(that could get to an IQ of even 190 as you claim) allow someone much dumber probably, that controls the AI to steal your job. Why wouldnt you counter that with an AI of your own, and remain competitive on the market using that tool for your own advantage. Why wouldnt you, a genius person, use such a tool to actually dominate on the market, found your own AI money making business, and actually adapt to this new circumstance. That is what a smart person would do, right?
Yes but that would require having access to cutting edge AI in order to dominate the other AI systems, which would require significant resources. At the moment i still have a significant edge over AI so for now i am focused on maximizing that opportunity
Great to see you touching on this topic my friend. I agree. AI is going to revolutionize the world in no small way. May we hope that AI benefits us all in health and finance.
Within 1 or 2 decades physical robots equipped with advanced AI softwares will be mass produced (There is already contemporary efforts currently by china and tesla to mass produce humanoid robots) Money will never become useless, even in the case we reach a state of complete abundance and we implement UBS (Universal basic services) and UBI (universal basic income) some people will want more stuff, they will have to get it through some kind of currency or credit unless it it made illegal to have more than others
@@InfiniteIQ Robots will surpass us for sure, but to have the ability to creatively think. Humans dont even understand what that means, never mind programming this into a robot. That will always be the distinguishing difference between robots and humans. As for Electricians there is far too much unpredictability and awkward motor movements involved in being an electrician. I have to disagree on this one.
hi i got question about quad n back lvls does certain levels have amount of iq that can be associatied with them? like what is normal iq for someone who is on quad 6? 3?4? or any lvl
What are you going to do with all that money you would save up? Sounds more reasonable to use that for solar powering, water gathering stations and all that. If everybody is out of a job, then money wont matter.
AI has already plateaued. Only minor improvements will emerge, less like to make mistakes. AI may be the perfect human but it won't pass human intelligence. AI might have 200 IQ but it will never reach crazy exponetial intelligence like 5000 IQ as some might suggest.
Firstly, IQ is only well defined in relation to the human population. Secondly, a revolutionary architecture could lead to improvements, and GPU power is still increasing exponentially. And finally, I share your sentiment. AI is overrated. In fact, investment firms already seem to be preparing for the bubble to burst.
I am the highest IQ in the world and according to me, making money is already impossible in terms of not being worth the effort. so i will keep increasing my IQ/knowledge. btw look it up: only 1% of humans worldwide are actually millionaires, so this confirms my thesis of making money being impossible, also because we have an efficient market hypothesis by Eugene Fama, no opportunities, too much competition and saturation. in my opinion AI is already being used to find and eliminate all business opportunities, and even if its not AI, humans are doing it. we are living in an dystopian nightmarish, kobayashi maru no win scenario, Daniel.
Lmao wdym 😂 If you work, you'll win. You work every day during years you can achieve whatever you want. Stop being delusional and thinking that IQ is best than anything. If you don't do anything of your life, you're just another average person, like 99% of population.
My goal is to deepen my relationship with the Ultimate source of all wealth, health, prosperity, intelligence, protection, love, understanding, guidance, and provision. God Himself. Without Him, no amount of intelligence or wealth will be of any consequence against AI.
Well I basically said that Id love to debate you regarding ai. I wholeheartedly dissagree with your take. I am studing an ai library called tinygrad. I believe have a good feeling of where ai can improve and where it is headed. You seem to think that ai will just break loose and destroy everything you love. But we already have autonomos intelligent beings walking around and we also have larger more intelligent entintes that we interact with daily: companies. If you wonder what a AGI will do, just look at what companies do. AGI will just be equivalent to companies getting more manpower. I have a couple of more in the chamber. I think we would have a great debate!
I'm 15 myself but I advice you to not take any claim these people make with grant certainty, no one can predict the future with such insurance, even if AI does reach this hypothetical state it's actual impact in general society and the ability to generate wealth is up to debate...we will just have to see, this stuff will obviously be at some point regulated by the government, that is for sure. And then again, the kind of artificial intelligence this man is talking about is not an LLM, it is often referred to as an "Agent" which is autonomous and has the ability to fully operate on its own, those AI agents have not even been released yet
@@A.MM661 yea i had pretty much came to the same conclusion myself. The last few years have proven that the advancement of AI isnt something that can be readily predicted and this guy seems to have a pretty pessimistic take on its forward trajectory. Anyway thanks judge holden
Not f$%ked, just unlikely to be able to contribute anything of value to the world, like most people. There are still skills you can work on that will always be valuable, such as communication ability and charisma
@@InfiniteIQ Thats fine by me, im not entirely concerned with contributing anything of value to the world and charisma/communication skills are my strong suit anyway
I think increasing your iq to 190 is infinitely more triumphing than becoming rich. Why not do something everyone thought was impossible? Why not be the first to achieve something so great it’s revolutionary? Why focus on silly money that everyone does everyday. Become the exception!!!!! Become remembered and known through history! Give the world hope in becoming smarter! We are rooting for you!
I plan on doing both
190 IQ will be terrible in reality
It will nullify your social skills and your ability to experience and enjoy life as a whole
@@PhilosophicalNonsense-wy9gy ok dimwit
@@PhilosophicalNonsense-wy9gyassumption assumption which you hav zero experience of
If the scenario you are outlining is becoming true, money will become obsolete.
Very long term probably, but that is a lot further away
I study software engineering, have created machine learning models, have been to conferences discussing the topics and I have talked with experts on the topic. No, AI will not replace all jobs in the next 5 years, especially jobs requiring high IQs. AI knows a lot but is not intelligent. AI is kinda dogshit compared to us when you think about it. ChatGpt needed 30M$ of electricity to get trained, I even think thats a lowball, and cant solve a lot of logic problems that most humans could solve. AI consumes staggering amounts of electricity and requires a lot of hardware only to be worse than a bee at image detection. Thats the situation as of now and if you dig deep, as maikeru01 said, its just numbers it only gets fed numbers and outputs a predictable outcome from those numbers, theres no logic there. It knows a lot but cant innovate, it cant invent new stuff because that new stuff would need to be fed into him in the first place for it to know about it. Even if you dont believe me on that, just imagine it does start to steal our intellectual power. Dont you think governments would step in? Dont you think we can just cut the alimentation from those AIs and thats it, it cant do shit anymore? There would need to be major fuck ups and loss of our inate need to defend our natural intelligence for AI to take over the world and control humans as some of you guys said.
As i said in the video, AI does not need to be in control for this to play out, super intelligent AI controlled by humans would result in the same outcome
@@InfiniteIQ Yeah and I mention that governments would crack down massively ont this if it was the case. Even if this doesnt happen and companies are allowed to lay off all their workers and everyone is now replaced by AI, companies will still need to sell their products, the products will still be there, the economy will adjust and if we produce so much with ai with little human involvment we will be in an economy where people barely need to work to survive, wich is also a good thing. Its a trchnological revolution just like the internet wich automated a lot of things, markets will adjust.
@@InfiniteIQ Not even the experts know what will actually happen.
@@InfiniteIQ I'm studying computer science and have my undergraduate project based on AI and it's really underwhelming how simple the task is compared to how complex building it is. But I agree, it's like the invention of the gun is to war, as AI is to business. If you can get more information, and transform that into knowledge in the business, the first person with the gun will have a gigantic advantage like Americans did in early history. They will most likely get a boom. But to be fair as soon as businesses get AI to be at this point, competitors will follow but be at a major disadvantage. This is basically data science.
@@InfiniteIQ I've had similar concerns in the past. My IQ was naturally 157 before I had a brain injury, and I'm about one and a half years off from recovering from that. I've not yet taken stock of exactly what I've lost, or where I am not, as I want to 'feel' whole before I measure where I'm at again yet I digress -- AI should shrink money making opportunities for people of this level, but it's not going to be apocalyptic. AI in this sense fundamentally becomes a tool. You mention in your videos that having high IQ has given you an abundance of good ideas, and "High-definition vision." I had both of these things before my injury -- the former has somewhat returned, but the latter is pretty much gone. Being where I am now, remembering how i was at my lowest, and at my peak before all of this, I can confidently say that anyone in my former shoes, the 157 IQ ones, would be able to use the 'tool' of AI, or even just formulate their own organic ideas into concrete plans to execute far better than someone of a lower IQ using AI as a tool. AI can expand the breadth of what someone like this can achieve, but depth is a hard ceiling -- AI can feed someone as much brilliance as it can, but if they cannot cognitively digest it, any product they output is going to be at best, inefficient, at worst, totally broken. Check out a lot of the AI trash on youtube right now for example, but imagine that extrapolated to every nook and cranny in a modern economy. The bigger concern is rather governments shutting out new opportunities but whatever -- point is, have faith in yourself. If I could be how I was again, I would not have a worry in the world. Everything was easy, and I felt comfortably isolated at the top of a pyramid.
AI is a scam. It has no intelligence. It is difficult for me to articulate what AI is but I'm very sure about this. When AI generates something, it does not have any understanding of what it is doing. For years it was outputting random images and humans would tell it which one was better. It shows you one cat with 4 legs and one with 6 legs and you choose the one with 4. And it removes the idea of cats having 6 legs from it's data base. But really it doesn't know the car has 4 legs. To a computer program an image is just a series of pixels. Not even pixels but code for pixels. And it has lists of patterns of pixels that humans have said are good and lists of patterns of pixels that humans have said are good. Same for audio, same for text. When you ask chat gpt a question or doesn't understand your question. It is replying with something generated to convince you that it was written by a human. It was designed to pass the turing test not to give accurate information. So you can ask it to give you code for something and it has a data base where it sees that code looks like this. Code starts with a left carrot and then it has these patterns of sentences and these file extensions and it's formatted like this. It will give you something that looks like code but probably doesn't work. You can ask it all kinds of things and unless it's really simple it can't give you any good answers because it isn't thinking. All it's doing is scouring it's server for replies that look like they are supposed to follow your message. That isn't intelligence. It does not have the potential to improve. It will not ever do anything beyond this. These companies are scams to trick dumb investors into giving their money to Nvidia, Google, Twitter, openai, etc. if you have stock you should probably dump it before the crash.
AI is still a big LLM...
That's AI rn
most people saying that kind of stuff fail to understand that your brain is basically the same, but infinitely more complex.
as the AI shit gets more complex (A LOT MORE THAN WHAT WE HAVE RN, LIKE AN INSANE LEVEL MORE), these sorts of arguments will lose ground.
of course, that might happen in 2 years or 200 years. who knows
AI is just probability where every node is a floating point number, so instead of definite output (This is a dog pic) it will see input of a different dog pic and approximate the result. So you can have a different angle or breed of dog and the AI will still know if its a dog. So a dog from a slightly different angle is less than 70% a "dog"
It's never fully sure or not fully sure it's just a rough guess, from 0% to 100% certainty
It's not really a thinking person, it just takes in a noise texture (random pixels) and moves the pixels into the right place to roughly approximate a dog (probability) It does the same thing for chat. A "dog" is just a large database of dog pics that are recorded as numbers. if you get a paper and drew some lines, the AI would look at the lines and see if it is at least 20% of the lines that is supposed to be shaped like the dogs in the database.
It will think "whats the likely output for" h e l l o. Otherwise it will say the same thing over and over like it does.
All the AI sees is random letters (Goodbye) and looks for the likely letters to respond. The AI may give you the wrong answers like fake names and places, because it's supposed to produce the most likely answer for the letters you provided.
like writing the number (6) that not pixel accurate it will still know but if you write an (s) it will know it's not a (6)
==============
With character chats, the AI has info of words and sentences that personality said in it's training data. And it has respond with letters that are a rough approximate of the words used in it's training data for "aggressive"
Like saying "Hello! i am the banana man!"
Take a look at GPT-o1 this shows ability to think about tasks
@@LOLSflint i think the difference is a human can run into a new situation that is similar but not the same as a past situation and come up with an idea for it. or a human can say something that they never heard someone else say before. the ai actually can't. if you ask it how many spark plugs a Lexus ES350 has, it will look through its database. if somewhere in thet data base it has an instance of someone asking this questions before, like maybe someone asked on reddit and a few people answered and said 6, then it can reply that it has 6. but if you ask it something that it doesn't have the info for, like if the car has any PFAS material in it, it will either say it doesn't know or it will give you an answer that it thinks sounds like a human answer. It may very confidently tell you that the car does not have any PFAS materials because it saw other instances of people asking that question about other products, but there is no link between whether the car has it or not and what the AI tells you. And there is no way for the AI to find that information, and the AI is not programmed and does not have the ability to simply connect to the internet and look for the owners manual or something like that. Whatever data is on it's server is what it has access to. It doesn't know if it's lying or not. It doesn't know if it's right or wrong. It's just generating the best reply it can. It will not be able to find patterns in sports betting or day trading. And it will not improve it's own software because that isn't what it does and that isn't what it can do.
I've accepted my future as a serf under capitalism. I'm just gonna try to spread good karma in hopes that we will eventually reach an enlightenment after the techno-feudalist dark ages.
Also please share your thoughts on Lion’s mane mushrooms there are many videos saying that it will increase your brain neurons. I figured that would help with increasing iq since memory is improved. Perhaps cover it in a futur iq diet video?
I mentioned my lions mane experience in my nootropics video
@@InfiniteIQI watched your video and you said you’d be interested in taking it again for a longer period of time to see if your body would adjust to the higher number of neurons that you hypothesized. Have you gotten back to it?
@@4unoahthere could be rewiring and stuff, but no new neurons are formed in humans and apes after birth, except in the hippocampus which is long term memory and doesn't have logic gates. I haven't watched his video but I'd confidently bet money you misunderstood him.
@@somerandomboi8239 Most scientists are in the danger-zone IQ range said to be between 130-150, if I am correct. Therein lies the problem.
Requisite variety and kolmogorov or descriptive programm complexity are key to understanding why we won't be able to achieve technological singularity. The system/programm cannot increase its complexity on its own. Complexity is the closest notion to intelligence as we understand it. Thus, we can likely not increase the intelligence of machine in an automated way. We can make them more and more complex but we cannot build programms that increase their own complexity. It is incomputable.
Disagree completely, it is efficiency of neural connections that gives rise to intelligence, not complexity
OK, but then you disagree with general statements made by Stephen wolfram who analysed these networks thoroughly. The conclusion is overwhelmingly speaking against the possibility of neural networks in the current form to handle irreducible problems. Neuro symbolic computation may resolve it to some extend. The core idea still remains. These devices cannot possibly surpass the laws imposed by Turing machines as they are build on top of them. Problem number 2: we can easily state in fuzzy everyday language things such as efficiency of neural networks without being able to provide a correspondence/isomorphism to respective low level Turing machines. Neural computing is still based on the classical paradigm. They can't surpass the computational limits imposed by the halting problem. You would have to give a formal explanation what this efficiency does in the context of computability? Is neural efficiency not predicated by the amount of neurons? Would more neurons not correspond to higher variety and its encoding require higher descriptive complexity. Could you build an efficient neural connection based on just a few neurons or a thousand? As far as I remember, more than a billion weights, respectively connection's were required for chatgpt's functionality. Pretty high variety if you ask me! The law of requisite variety and its related notion of incomputality just make the claim that once you designed the algorithm and run it you can't make it more complex. In order to do so you would have to jncrease its number of neurons possibly needing more storage. This process can be infinitely extended by humans. Once you start the device however at the level it is , it can't become more complex. More specialised and refined or adapted through training. That is possible. However, not more complex. You would have to provide a formal proof of things being different. Cybernetics and theory of computation are the only valid theories so far we have for describing computational processes. Everything else are speculations enabled by the fuzzy nature of natural language. More than 2000 years of lack of results in philosophy showcase this. We can't go very far with ordinary language. The major reason why chatgpt seems so fantastic. Let's phrase it differently. It's easy to talk the talk and difficult to walk the walk. -'The map is not the territory'
I get the idea that you are referring here to speed. TRUE THAT. However, without sufficient variety or complexity you cantvdo much to begin with. A simple neural network may be very efficient if run on a photonic computer but it can only perform the operation 1 + 1 quasi infinitely faster.
@@InfiniteIQ Efficiency is a key attribute of intelligence, involving performing tasks faster or with minimal computation.
Irreducible systems (e.g., chaotic systems, P-vs-NP conjecture) pose limits to computational reductions.
Computational irreducibility suggests some tasks cannot be simplified.
Efficiency is more applicable in reducible/regular tasks and complex tasks with regularities.
Building efficient neural networks can reduce neuron usage, but cannot go below the minimum variety needed to compute a problem.
Using a single neuron for efficient computation is limited in certain contexts.
While efficiency fits within IQ research, it’s insufficient for rigorous computational definitions of intelligence.
Precise definitions of intelligence remain elusive; current notions are ambiguous.
Ross Ashby’s definition of intelligence highlights the balance between efficiency and variety in neural networks.
Specialization for efficiency can lead to a loss of adaptability in systems.
@InfiniteIQ I agree that AI will largely replace humans in many areas. We can automate most, if not all, things we have invented and formalized properly. However, we cannot automate the process of automation itself, as discussed earlier. Over time, AI will likely replace a significant number of people. Remember, AI is the world's best copycat-far superior to us. Even if humans come up with new practices, they will need to improve efficiency, and AI will likely perform that task better. This doesn't mean humans have no place, but it does mean that most of us will be replaced unless we pursue becoming renaissance individuals, which most won’t. It also raises questions about the kind of world that will emerge once AI is fully integrated. If wealth is not distributed adequately, we could again face a scenario where a small group holds much, and the majority has little. The idea of AI replacing humans due to efficiency gains is widely accepted, but it’s a short-sighted view when considering long-term economic stability. Unfortunately, as beings who learn through trial and error, we may drift in that direction.
Hey man, I know your goal has changed but it'd be cool to know when you're planning to take the IQ test again. Probably after quad 9 back after your 2 year money making period. If you're planning to not take it too many times, that's also fair. If it were me and I was doing an IQ raising project I'd like to measure that as frequently as possible without ruining the results.
IQ tests can only accurately measure up to 160 so there is little point testing when i am already 150+ but i will test again when i hit quad 9 back to see if i am at 160 at that time
@@InfiniteIQ There are some IQ tests capable of measuring above that. Try doing the sigma test light by Sigma Society. Its an ultra high IQ community created by a brazilian genius with an IQ of 233. And please, reply to some of my comments towards you in other videos, there is information that I believe you will find interesting.
Hi Daniel, I have a question. Can I increase my IQ infinitely with Quad N-Back? Or is there an IQ limit that I can reach with the Quad N-Back?
The gains are near infinite, but they slow down a lot as you progress
Its like muscle Training. At first your gains are big and as time progress it shrinks
@@mrxwalker4479 That's true
Ai wont replace construction any time soon. I think it takes more time than five years for AI to implement in our economy. And you can do thousands of Iq tests and improve your iq. But that doesn't mean per se you become more intelligent. Because if you study iq tests, you mostly learn the similarities among different iq tests.
So what do I do to increase my iq
@@RedditTales_009 you cannot.
I would like to understand what you even mean by IQ?
Could you exponentially increase your knowledge and skills through memorization in specific fields?
Yes of course, through different methodologies you could increase your knowledge and also expand your perspective and way of thinking, but if you by IQ mean actual processing intelligence that is mostly genetic, meaning it depends upon your genetic material, some people will always be inherently more intelligent than you, they surely will not be as knowledgeable but more intelligent for sure.
Yup, you can become better in IQ test tasks, doesn't mean your brain gets better beyond that.
@@RedditTales_009 you cannot really
improve diet / supplements, improve thinking skills (eg feuerstein method), increase knowledge
I'd advise to specialize knowledge-wise instead
I suggest you watch my earlier videos for how to increase your intelligence
you're not smart, but trust me you are a great person
Edit : no one knows intelligence , even me
Its too complex
I'll take that
He does seem smart. You don't know what smart looks like.
@@camharris1664 what is the definition of smart
Dude bye. We're not having this conversation.
@@camharris1664 alright, mister smart
I agree with you, we have still not realized how soon this change is coming. AI's invention is akin to the discovery of fire, drastic changes will follow suit and we need to prepare ourselves for that.
"Prepare" in what way and to what degree can you prepare for such an impactful hypothetical future AI will have?
Any degree of preparedness will be barely relevant compared to someone entirely unprepared when it comes to being effected by the future aspects of AI.
If AI brings upon a golden age of technological and scientific advancement solving global crisis and achieving abundance of resources we can expect the level of lifestyle to grow in quality, if it takes the polar opposite direction we could live in a super intelligence police state in which we will always be outsmarted by a 10,000 IQ super intelligent entity.
All we can do is see it unfold, the general population even if aware would not be able to do much to prepare.
The only way to prepare is to increase our own intelligence and acquire resources (money) but yes all preparation will likely be futile
Thank you,Daniel.Please also make a video about best skills to learn during AI era? I mean,certain people also will get rich during AI era and it is necessary to learn skills that will be useful during that time.Please make a video about it that we won't miss the train when it comes
This is interesting
I'll make a video about that soon
Start a business
Hi from Brazil! I just want to say that you have a special channel for me
Same. Check my other comment in this section.
AI will not reach super intelligence, or even general intelligence, anytime soon. I almost want to go as far as to say it's even an impossibility to achieve either of those things period and that we have made a categorical error.
wrong
Focusing on making more money is great, but it would be even better if you created a second channel dedicated to business insights and strategies-maybe call it 'Infinite Money.' However, since this channel is named 'Infinite IQ,' we expect content that delivers on that promise. We'd love to hear more about techniques and methods to boost IQ, and we would deeply appreciate regular, actionable tips on enhancing cognitive abilities. Keep the content aligned with the name, and continue sharing new ways we can improve our intelligence.
I am not here to serve anyone, i am not interested in growing on TH-cam, it is worthless to me. I just enjoy sharing my ideas and what i am working on and learning, and if people get value from that then that is a bonus
Because ai will as you said be controlled by people and will not be able to develop true agency on its own there will still be lots of economic inefficiencies. What will occur is actually the average man will be able to access ai more in the future, and will be able to develop businesses a lot faster. Intelligence will be the factor for developing the ideas which you then have an ai quickly do the technical work for. We already see this occurring with complex models being developed by chat gpt in firms which might have taken them months to formalize on their own. It can create a golden age of opportunity for the middle class and arguably already has.
Those with more resources will be able to use more powerful AI tools to outcompete those with less resources
@@InfiniteIQ supercomputers will have competition from geopolitical competitors and moore's law will over time make the tech viable for civilians to use, similar to how 1960s supercomputers are less powerful than your smartphone
It’s impossible for the secnrio to happen unless ai is independent
Companies relay on employees to get paid if all companies have no coustmers(employees)
then who will pay them if all companies use AI
Thus
Ai will not replace humans unless its self driven then that will ofc be proplem
Companies relay on employees to get paid if all companies have no coustmers then who will pay them
Thus
Ai will not replace humans unless its self driven then that will ofc be proplem
human controlled AI will result in the same outcome
@@InfiniteIQ I don’t think so i think it will happen but it will make the system collapse because companies relying on employees to gain sales and employees relay on companies to gain salaries if any of both sides fall both do so
Если ИИ будет настолько хорош, ему не составит труда уравнять людей которые разбогатеют к тому моменту в пользу своих разработчиков и правительства. Надеяться что мы будем играть по тем же правилам в эпоху ИИ что и сейчас это наивно на мой взгляд. Это просто будет новая реальность как после падения астероида.
At some point. First of all, the rich and powerful will do all they can to stay rich and powerful. The unemployed will get UBI to stay quiet.
Do u do other brain training other than quad n back? I mean quad n back trains certain types of intelligence (namely/mainly working memory capacity) but there is more aspects
All areas of intelligence increased from the training, working memory appears to be the bottleneck for all forms of intelligence
Are you able to provide some of the resources you've been looking at? I'd like to do some deep dives.
David Shapiro on TH-cam is a good place to start
@@InfiniteIQ As someone well versed in CS & AI, that guy is IMHO a charlatan basically making a cult in order to gain money from his followers.
This comment is unrelated to the video, but I was wondering when the part 2 of your “38 books that changed my life” video will come out.
What if we were to create paper markets? Like literally, go back to basics to stop ai from taking advantage of the algorithm on a computer? Would that not work?
In theory yes, de-evolution is not the answer though
Would you give E-commerce a shot? There are a lot of Ecom players out there and I wonder how you might approach it?
I'm not interested in ecom, there is infinite competition
What should I get into then? I’ve been thinking about learning trading and algo-trading after that, which AI will surely make much more difficult than it already is, but I know some algo traders and it can be highly profitable once (& if) you find an edge which takes a couple of years usually.
The sports betting thing seems interesting, but it seems to me like its just gambling.
I’m 16 and broke so no money to start with and parents didn’t allow me to get a summer job
My goal is 10k/month before 25
I suggest finding a business with a real world component (offline), they will survive the longest. I shared one business idea you can copy recently
Get into GOD
What happened for you to go from achieving quad 9 back in your last video a week ago to this?
That is still my goal, it is my #2 goal
Im also on sports betting, if you want to share ideas let me know.
Do you recommend starting with Dual-N-Back before jumping straight into Quad-N-Back? What's your recommendation on how to start?
Start with Quad
@@InfiniteIQ Daniel, I hope you are doing well. I have a request: can you elaborate on the words you spoke in the previous video about achieving means of living off of passive income in 1.5/2 years? I wish you would make a video about how you achieved that. I greatly appreciate your channel since our goals are identical. Please, excuse my errors of grammar.
Well, with super intelligent AI there could be many scenarios. Lets take the hypothetical scenario where AI becomes smarter than a human is controlled by a human, and serves as a very powerful money making tool. Even though its less likely that something actually smarter than a human would want to remain subsurvient to humans, but lets assume its possible. Why would that make you change your number one priority of becoming more intelligent into a goal of making more money? Why would a super intelligent person like you(that could get to an IQ of even 190 as you claim) allow someone much dumber probably, that controls the AI to steal your job. Why wouldnt you counter that with an AI of your own, and remain competitive on the market using that tool for your own advantage. Why wouldnt you, a genius person, use such a tool to actually dominate on the market, found your own AI money making business, and actually adapt to this new circumstance. That is what a smart person would do, right?
Yes but that would require having access to cutting edge AI in order to dominate the other AI systems, which would require significant resources. At the moment i still have a significant edge over AI so for now i am focused on maximizing that opportunity
Great to see you touching on this topic my friend. I agree. AI is going to revolutionize the world in no small way. May we hope that AI benefits us all in health and finance.
why are you against imagestreaming it works better than quad n-back?
It works on only a small subset of the population
@@InfiniteIQ no it doesn't it worked wonders for me
What part of small subset you didn't understand genius ? @@elena_m19
@@elena_m19 doesn't disprove his point lmao? you can be a small subset of population
how much did it raise your fluid intelligence?
I do 30min of QuadNBack daily, is it ok?
Yes that's good
Infinite iq I have a question do you have a religion or believe in anything?
No, but i like the idea of Objectivism
How long until 1. you don't need money to live an otherwise standard middle-class life and 2. money becomes essentially useless?
It will be sooner than we think I imagine. Unprecidented change
1. 5-10 years, once humanoids are mass production. 2. Money will always be useful if you want anything other than mediocrity
Within 1 or 2 decades physical robots equipped with advanced AI softwares will be mass produced (There is already contemporary efforts currently by china and tesla to mass produce humanoid robots)
Money will never become useless, even in the case we reach a state of complete abundance and we implement UBS (Universal basic services) and UBI (universal basic income) some people will want more stuff, they will have to get it through some kind of currency or credit unless it it made illegal to have more than others
idk how i found. this video, but damn i believe this smh
Man like Joseph Castello 😂
You need to learn skills that AI are unlikely to take in our lifetime such as a trade like an electrician or creative writing.
creative writing will be entirely done by AI within a year or so, and humanoids will take all electrician roles within 10-15 years
@@InfiniteIQ Robots will surpass us for sure, but to have the ability to creatively think. Humans dont even understand what that means, never mind programming this into a robot. That will always be the distinguishing difference between robots and humans.
As for Electricians there is far too much unpredictability and awkward motor movements involved in being an electrician. I have to disagree on this one.
hi i got question about quad n back lvls does certain levels have amount of iq that can be associatied with them? like what is normal iq for someone who is on quad 6? 3?4? or any lvl
there's always rage baiting and fear mongering... that seems to be doing good for a lot of people...
Not fear mongering, just sharing proactive steps I'm taking
It just makes me want to live in the forest
So IQ will be meaningless in 2-5 years?
Not meaningless, but far less useful
What are you going to do with all that money you would save up? Sounds more reasonable to use that for solar powering, water gathering stations and all that. If everybody is out of a job, then money wont matter.
Money will always buy more freedom
@@InfiniteIQ Yeah been like that always. Will it be like that in 2-5 years though?
Good video, are you suggesting in this case that we should stop studying and start making money now?
Yes, if money is your goal then study should be limited to areas that can generate a quick return
AI has already plateaued. Only minor improvements will emerge, less like to make mistakes. AI may be the perfect human but it won't pass human intelligence. AI might have 200 IQ but it will never reach crazy exponetial intelligence like 5000 IQ as some might suggest.
I urge you to conduct more research, respectfully 🙂
At minimum it will be more intelligent than the combination of the best aspects of intelligence of the smartest people in the world.
Firstly, IQ is only well defined in relation to the human population. Secondly, a revolutionary architecture could lead to improvements, and GPU power is still increasing exponentially. And finally, I share your sentiment. AI is overrated. In fact, investment firms already seem to be preparing for the bubble to burst.
@@giladkingsley That would be the maximum imo. I really dont see this becoming as big as most people think
famous last words, andreas pilates-patsalidis
I watch you from algeria
Why would you or other profitable traders/businessmen be unprofitable after 2-5 years?
Super intelligent AI will create infinite competition, eroding profitability for opportunities to near zero
I am the highest IQ in the world and according to me, making money is already impossible in terms of not being worth the effort. so i will keep increasing my IQ/knowledge. btw look it up: only 1% of humans worldwide are actually millionaires, so this confirms my thesis of making money being impossible, also because we have an efficient market hypothesis by Eugene Fama, no opportunities, too much competition and saturation. in my opinion AI is already being used to find and eliminate all business opportunities, and even if its not AI, humans are doing it. we are living in an dystopian nightmarish, kobayashi maru no win scenario, Daniel.
"I am the highest IQ in the world and according to me".... What? Lmfao. look up dunning kruger effect.
@@fish-rm6xl the author of this video is also violating dunning kruger
Lmao wdym 😂 If you work, you'll win. You work every day during years you can achieve whatever you want. Stop being delusional and thinking that IQ is best than anything. If you don't do anything of your life, you're just another average person, like 99% of population.
@@Adam-cn5ib he definitely doesnt strike me as 190. 😹
@@fish-rm6xl nah he's not 190, but he legit tested at 149 I believe (not an internet test, it was with a proctor)
My goal is to deepen my relationship with the Ultimate source of all wealth, health, prosperity, intelligence, protection, love, understanding, guidance, and provision. God Himself. Without Him, no amount of intelligence or wealth will be of any consequence against AI.
hey man why did you delete my commnent
I don't delete comments, TH-cam deletes some automatically sometimes
Well I basically said that Id love to debate you regarding ai. I wholeheartedly dissagree with your take. I am studing an ai library called tinygrad. I believe have a good feeling of where ai can improve and where it is headed.
You seem to think that ai will just break loose and destroy everything you love. But we already have autonomos intelligent beings walking around and we also have larger more intelligent entintes that we interact with daily: companies.
If you wonder what a AGI will do, just look at what companies do. AGI will just be equivalent to companies getting more manpower.
I have a couple of more in the chamber. I think we would have a great debate!
My thoughts too
Can you make a video on how you jumped into business
As soon as you said sports betting and trading I knew this was bs.
My ideas may be flawed, but they are shared honestly without motive
Im 16 right now and im still in school, by your logic am i just completely fucked then ?
Yes.
I'm 15 myself but I advice you to not take any claim these people make with grant certainty, no one can predict the future with such insurance, even if AI does reach this hypothetical state it's actual impact in general society and the ability to generate wealth is up to debate...we will just have to see, this stuff will obviously be at some point regulated by the government, that is for sure.
And then again, the kind of artificial intelligence this man is talking about is not an LLM, it is often referred to as an "Agent" which is autonomous and has the ability to fully operate on its own, those AI agents have not even been released yet
@@A.MM661 yea i had pretty much came to the same conclusion myself. The last few years have proven that the advancement of AI isnt something that can be readily predicted and this guy seems to have a pretty pessimistic take on its forward trajectory. Anyway thanks judge holden
Not f$%ked, just unlikely to be able to contribute anything of value to the world, like most people. There are still skills you can work on that will always be valuable, such as communication ability and charisma
@@InfiniteIQ Thats fine by me, im not entirely concerned with contributing anything of value to the world and charisma/communication skills are my strong suit anyway
would there be AI stock trading bots in 15 years making a killing? so far their useless
Lol we gonna have AGI in 5 years and youre asking specifically about trading bots in 15 years
@@quantumspark343 We won't have it even in 50 years
@@FNproject lol ok sure
This ☝️ spamming click bate is awful