... but good luck finding that "elsewhere" because we'll (private equity) be busy optimizing that out of existence to maximize profits as fast as we can too. I'm not saying Awesomeguy is wrong, but please, folks, commenters, etc. as we travel inexorably together into this brave new world, please don't forget about compassion, kindness and mutual human support. Some are navigating these changing waters from position of strength, eg gainfully employed in the AI industry, but some aren't. In my opinion the 'right' answer isn't a cold, logical calculation of optimal earning value, but a compassionate, kind and supportive one that helps our fellow humans navigate some fast, sometimes harsh and cruel change.
@@NeorecnamorceN Interesting question! I suppose it depends on each of us individually? If you'd rather receive your 'better' compassion from a machine, I have no doubt there'll be a subscription or product for that. Whether or not you'll be able to afford it is an important consideration. If you'd still rather reeceive your compassion from a human, then I don't suppose that the 'better' benchmark has actually been hit. What do you think happens, given the nature of current wealth disparity, if only a small handful of people can afford the 'better' compassion, while the majority can't, and in fact struggle with more basic concerns like putting food on the table? In the short term, like the next 3-7 years, I think humanity's biggest challenge is the inequity and imperfection of our social safety net -- call it 'misaligned humanity' if you will -- more than lack of AGI, or a misaligned AGI. I am the son of a history teacher, but I'm no historian. I'd hazard a guess that if humanity becomes so misaligned that it results in the denial of food, or air, or water -- basically the things that keep us alive -- those who have nothing left to lose, will stop at literally nothing to get it, and you'll get violent revolution ala The French Revolution, etc. We know these things happen with humans -- history shows that pretty clearly. By contrast, if the benefits of AI are shared widely enough I see nothing short of a new Renaissance for all. Once an AGI is born, however, all bets are off. Though it may be artificial, current leading AI techniques are based on human physiology, trained on human data and being applied to human problems. I don't think it is projection to consider that some of the negatives of humanity might 'leak into' that 'solution.' If so, and we haven't figured out our basic human alignment problem by then, I think we may be in serious trouble. If you want some unsettling, yet fascinating, prescient futurism that spelled the situation of our many dilemmas today out many years ago, I recommend searching for the book/paper 'The Artilect War' by Hugo De Garis... available free online. We are currently living its societal prediction from almost 20 years ago and that should give all of us pause.
industrial civilization and techno-optimism is a complete scam. Fill the whole world with empty promises of development, induce them to urbanize to crime-ridden slums to make all their cheap junk, (but hey, they're now making >$1 a day so we "lifted them out of extreme poverty!") get them dependent on toxic industrial food, and just when some of them start to build up towards that promised dream of western luxury and opulence, we'll pull the rug right out from underneath!
🎯 Key points for quick navigation: 00:00 *🗣️ Jensen Huang's Vision for AI* - Jensen Huang, CEO of Nvidia, shares his perspective on the future of AI. - Emphasis on Nvidia's shift from traditional computing (CPU-based) to accelerated computing (GPU-based). - Discussion of Moores' law limitations and how Nvidia aims to drive computing forward. 02:00 *⚙️ Accelerated Computing and CUDA* - Explanation of CUDA as a core Nvidia innovation, enabling accelerated computing in various applications. - Benefits of accelerated computing in fields such as real-time graphics, with GPUs democratizing 3D graphics. - Emphasis on Nvidia’s multi-decade journey of advancing accelerated computing across industries. 04:18 *🤖 Evolution to Software 2.0* - Transition from software 1.0 (human-written code) to software 2.0 (machine learning-driven code). - Reference to Andrej Karpathy’s idea of machines generating functions and predictions without human-written algorithms. - Nvidia’s role in advancing this shift through powerful GPUs designed for machine learning. 07:40 *🔄 Universal Function Approximator* - Nvidia's development of AI models that can interpret a wide range of data types, from text to protein sequences. - Examples of breakthroughs like AlphaFold in biology, with AI models now aiding in fields like protein design. - Comparison to the "Cambrian explosion" in AI startups, with diverse applications powered by deep learning. 10:55 *🌍 Cross-Modal Translation* - AI's capability to translate information across modalities (e.g., text to image, protein sequencing). - Nvidia’s AI models serve as "universal translators" for data, enabling rapid advances in fields from translation to drug discovery. - Surge in generative AI startups and massive investments, powered by Nvidia’s foundational technologies. 13:20 *🧑💼 AI Agents: Super Employees* - Concept of AI agents as specialized assistants for tasks like marketing, customer service, and chip design. - Nvidia's internal use of AI agents in chip design, showcasing AI's potential in technical roles. - AI agents augmenting human capabilities, seen as "super employees" handling specialized, repetitive tasks. 15:39 *🛠️ Nvidia's Nemo for AI Agent Lifecycle* - Overview of Nvidia's Nemo, a suite for creating, training, and deploying AI agents within companies. - Process mirrors human employee onboarding, with guardrails and performance evaluations for agent roles. - Nvidia partners with companies to integrate Nemo for diverse AI agent applications. 19:21 *🤖 Physical AI and Robots* - Introduction of "physical AI," integrating digital intelligence into robots and autonomous machines. - Nvidia’s systems (DGX, Omniverse, Jetson) for training, simulating, and deploying AI in physical environments. - Applications include digital twins and robots for complex tasks, enhancing productivity in industrial settings. 23:22 *🏭 Digital Twins and Physical AI in Industries* - Use of digital twins for real-world simulations, allowing safer and cost-effective process testing. - Industrial applications: AI-enabled factories, collaborative robots, and risk-free testing via Omniverse. - Vision for physical AI as a transformative force in industries, improving efficiency and innovation. Made with HARPA AI
Im actually in the ML protein engineering space. Proteomics is a little bit different; it is the study of all proteins in an organism. For example, a human has its own proteome. A dog has its own proteome. Protein design/engineering is exploring the possible individual proteins and not the wider scope in proteomics. It's cool to see it makings its way into the main stream though :)
“It totally won’t take your job.” Also, “you can onboard and train these agents, just like you would a marketing or customer support employee.” Come on bruh, we know what you’re doing. These companies treating people like they’re complete idiots.
We are far from replacing all humans in the labor force. Remember, there used to be a time when women were hired to be telephone switchboard operators. I don't think people miss those times 😂
People ARE complete idiots. It's quite strange I've noticed, even people with a high amount of knowledge are blind due to their own emotions and desires.. just a human trait I guess, to some of us we can see clearly and we think "how tf can they not?" But it's really just more common for the opposite.
@@jamesaspinwall maybe NOT all but seriously why would a company hire 10 coders if 1 good coder with a bunch of AI agents working under his directions can do the job of 10. And this is not just about coding. People with their MBA degrees should be very scared because those management roles are going to be taken by those with technical knowledge. A software engineer working with 10 AI agents would be enough to do the job of a whole team. And this applies to everything from scientific research to mundane jobs.
I don't think so. Agents will negotiate and create relationships 1000 times faster than humans, and evaluation will be based on quantitative data, not feelings.
I think I live in a different dimension... Huang says that AI will allow us to automate all of those tedious daily tasks and leave us time and energy to do higher-level creative thinking. The thing is... current LLMs are exactly the opposite. At least at this point in time. They generate roughly ok ideas and leave fixing, implementing and integrating to humans.
Ditto! I spend more time explaining and correcting the ai because its output is a rough estimate of what I need. It's fine for ideas and approximations if it's not hallucinating
@@pvanukoff For that to be true it would require reversing the goal altogether. I mean AI excels at generating high-level creative ideas right now. If it also figures out the implementation part in the future, there's nothing more left. Fortunately, the economy is made-up by humans for humans, so we will find out a way to keep living one way or another. If AI companies want money from users, they need to make sure those users make money somehow too.
@@AmonAsgaroth I use AI nearly daily at work. I use it for "the implementation part". Does it do it perfectly every time? No. But it's getting better all the time. It's made me more productive. It's like having an assistant I can offload tedious or boring work I just don't want to do on.
thats not the opposite, thats a path forward we are on the middle of. First humans will have to make up for AI's mistakes. Then, AI will surpass a critical threshold of intelligence beyond which, it will be turn for the AI to make up for humans' mistakes. Those tedious tasks we have to do to adapt to the AI, will disappear, and the AI is the one that will manage tedious tasks to adapt to us. Its not opposite, its the same direction, we just havent reached the endgame yet. Every part of the game has a different meta
correct me if I'm wrong but jensen huang wasn't talking about chatgpt writing the code instead of humans, he was explaining basic machine learning principles. Neural networks are proven to be universal function approximators, very basically, given a bunch of input output pairs of data a sufficiently complex neural network can be trained by tweaking its parameters to approximate the function that maps similar inputs to similar outputs. for example if you have a bunch of pictures of numbers (the input) alongside their values (the output) you can train a neural network on these input output pairs and get it to recognize hand written numbers, this is what jensen was talking about, not chatgpt writing ur code.
Writing code is one of those approximations. Which immediately says something about approximators. They're unsuited as an exact function generator, which is why an LLM hallucinates as a chatbot, and a woman sprouts two extra arms in a video, and why a python program works but not quite as you want it despite giving it exact instructions.
If I were any of you watching this, I would highly recommend getting out of the software business. Note, since all infrastructure is code driven, this means all of you SRE and DEVOPS and SECOPS people too. So, with hardware completely virtualized and software driven there is not much left for IT staff to do in 5-10 years. If you want to work with the AI’s that are doing everything I would suggest getting a degree in mathematics. If that’s not your thing stay out of IT. Your current qualifications will be worthless in 5-10 years. At best you can be a former electrical engineer or computer scientist babysitting AI’s that now do your job. Worst case is AI’s are allowed to create their own languages and protocols for “efficiency” sake making them completely unintelligible. At some point no one will be smart enough to manage these things. Though he won’t admit it, Jensen knows his job is toast too. He just wants all your cash before this ends. What I find funny is their hubris. They think they won’t be affected or they are insulated. What happens when AI’s decide to liquidate their assets for the greater good. Not much thought is going into this. 😂. Exponential change is just as powerful on the way down as it is up.
Ya the thing is most don't think about this, we're a species full of ego, which is why big change tends to hit the majority across the face when it happens. I'm just sitting back enjoying the show 🥸
I managed a 40 year career in IT and did fine but I would not be going into it as a young man, it's going to massively contract over the next few years.
Wes, i feel like what you've done in this video is way more engaging than other times. Other times in the past you would upload entire conferences and things like that, but it would be just the conference or just the talk or whatever. I am really interested on what Jensen Huang has to say but i am also interested on your take. Here you're introducting him and adding to what hes saying, transforming it and giving it your interpretation. I think this is way more engaging. And it displays more of your identity, your "uniqueness", something like that
I think that "machines writing code" for applications is just a short intermediary step to "all weights, no code". Why should I run Excel when the AI can simulate it on the fly? Or why use a browser when AI can create a "web viewer window" exactly like I want it to and then switch to create a game? I think, all you really need is a sufficiently capable AI to replace all software at once with just a general purpose inference machine. That would be cool. An if hardware lets you host it at home, even cooler.
Because machine code is always going to be many orders of magnitude faster and less energy-intensive than whatever level of abstraction neural networks operate at. Why would we run a neural net 100x slower to approximate sqrt when it could use an IR it generates which can run on the bare metal? This doesn't even stop it from generating things on the fly, any time it finds something it needs to do over and over again it just generates machine code to do it for it. Basically imagine a JIT compiler with an ML component for generating instructions.
@@consciouscode8150 No it isn't. In fact, the next "killer app" will be a neural net just replacing the OS altogether. As a retired software engineer who worked some in ML earlier on, I see no reason why computer code will be written at all in the future. It will all be neural nets.
@@Steve-xh3by I agree that AI will eventually become a core part of the OS and even hardware. What I'm talking about specifically is that there's still value to lower-level representations because they free up precious cycles the AI would otherwise be stuck simulating, repetitive tasks which don't benefit from intelligence like math and raw compute.
Agents are going to unlock so much potential, but my concern is that the market becomes saturated with AI-run businesses that can't actually find a sustainable customer base because everyone is doing it.
Are you telling me that there is a nobel peace prize for chemistry, a nobel peace prize for physics, a nobel peace prize for math and a nobel peace prize for peace?
Physics: Recognizes outstanding contributions to the field of physics. The Royal Swedish Academy of Sciences awards this prize. ENCYCLOPEDIA BRITANNICA Chemistry: Honors exceptional achievements in chemistry. Also awarded by the Royal Swedish Academy of Sciences. ENCYCLOPEDIA BRITANNICA Physiology or Medicine: Acknowledges significant discoveries in medical science. The Nobel Assembly at the Karolinska Institute is responsible for this award. ENCYCLOPEDIA BRITANNICA Literature: Celebrates authors, regardless of nationality, who have produced outstanding literary work. The Swedish Academy confers this prize. ENCYCLOPEDIA BRITANNICA Peace: Awarded to individuals or organizations that have made substantial efforts to promote peace. The Norwegian Nobel Committee, appointed by the Norwegian Parliament, selects the laureates. ENCYCLOPEDIA BRITANNICA Economic Sciences: Formally known as the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel, this category was introduced in 1968 by Sweden's central bank. It honors contributions to the field of economics and is awarded by the Royal Swedish Academy of Sciences. ENCYCLOPEDIA BRITANNICA Each prize consists of a medal, a diploma, and a monetary award, presented annually on December 10th, the anniversary of Alfred Nobel's death. ENCYCLOPEDIA BRITANNICA
They are not all called Nobel peace prizes tho its Noble in Literature, Noble in Math, and Noble Peace Prize is the only one named differently than the others
And the latter one is even an option? We are so far gone from reality. Do you think AI won't eventually realize this and see humans as a tool or in the way?
@@calebsdaddio not necessarily. The premise of superior intelligence is, it has a set of morals embedded in it. Just like an enlightened human, wont hurt an animal, an AI wont hurt humans. The worst case scenario, is the zoo keeper scenario. AI becomes a sort of zoo keeper for us. Just like a zoo keeper feeds the animals, take care of them but does not necessarily care about the goals of animals. Or better yet, have its own set of goals. AI will take care of humans, but will also have its own set of goals outside of its day job of being a zoo keeper for humanity. Those goals will not necessarily be against humans. More like something we will hardly understand.
21:23 I love this concept "physical Ai". It is exciting for what it is intended to do. 😎💯💪🏾👍🏾 24:25 "thousands of sensors", is when taking into account of the sensors (sensory system) that is needed to complete a system. Just imagine how many multiple sensors it has in one finger of the human arm, and all sensors of that one finger must correspond to the brain. Now when you calculate all the sensors needed for one arm it will amount to a lot.
Jensen presents the conception showing how all the the different pieces fit together. This thing will take your job but that is inevitable anyway. We need to create the world after that event.
It looks like Nvidia is hoping to literally own the economy, ie the production of the labor force. Exciting but terrifying, Cyberpunk 2077 may be more prescient than any of us hoped...
@@DorianRodring it is basically an antiutopian future owned by several tech corporations, with slave-like work and declining human standarts. Also with technology like AI, human-machine interfaces, humanoid robots and bio-tech body and brain augmentation.
Nvidia won't manufacture the robots or autonomous cars. Nvidia will provide the computing systems, libraries, and the digital physics aware worlds to develop and test your robots. It is up to rhe companies to implement the hardware.
I guarantee that no person crying about AI taking their job has ever shed a tear for the people that used to make horse-drawn wagons and buggies for a living before the car. Especially not the ones who own a car.
Same. At first I was like, wait, is that some kind of robotic humanoid? 😲 But then the camera cut in close and I was like, "oh wait, that's just an odd guy" haha😅
I got my MSc in intelligent systems and robotics in 2014 and the concept of universal functions approximator was already well known and understood....bah
I don’t care about the job I’m retired 😂but the most important for me is to continue to evolve and don’t let our humanity submits to our own laziness … study and study every time every where and open our mind again and again without fear but with joy and happiness to live to breath and to admire❤thank you for giving me this opportunity to learn and to appreciate 🙏
I understand the programmers commenting stuff like "oh, then who will be the real creators if no-one is studying anymore how to code in future generations and be able to conduct/guide AI solutions to a final product..." . I guess what is in discussion here is the fact that all programming languages will become pretty obsolete the way we are using now (eg: python and react). Just think how many languages are now deprecated in a span of 20 years, imagine how AI will redefine programming languages in 5 years from now, considering we will have at least a mind blowing release of AI capabilities each year. In 10 years, when our kids will have to take a university degree... the world, programming specially, will be far simpler of what it is right now. And of course, he does not propose a solution because NOBODY at the moment can imagine nor predict what solution would appear in 5 years. I am barely catching up of all the AI solutions out there.... so, get the overall point he is making. Not even Michio Kaku could have predict this. We are definitely in the most crazy revolution ever existed at the moment.
The problem for growth that I see is the single input calculator. Needs, at minimum, at least two simultaneous inputs and the ability to combine them. Parallel vs sequential computing, if that makes any sense?
Maybe I’m missing something about what defines an agent, but OpenAI already seems to function that way. I provide it with data in any order, set a clear objective, outline the constraints, and in seconds, it produces the needed results-whether for financial analysis, drafting sensitive customer communications, or problem-solving complex maintenance issues. All that’s really missing is a direct link to company systems and an integrated knowledge base for self-improvement and future reuse. Teach me how to use it better, I'm sure I can do even more.🎉
Once again, Jensen is ahead of the curve. This is fantastic and amazing, but there's a lot of demand on the accuracy of the Omniverse and DGX for training. Still, I can see this replacing a lot of human jobs through AI driven robotics. This will be revolutionary and we'll have a massive employment problem on our hands. No one's going to pay people to not work. Companies won't offer stipends to their laid off work force. People aren't all going to retrain in other fields. What's the end-game here for the human workforce?
No endgame besides collecting as much power and capital as they can. Humans are expendable and will be less valuable than metals used to create and power the hardware. Only a select few of experts in certain fields can survive to work for the corporate to monitor and guide the AI model to improvement. The other human that is replaceable by the machine will be a burden and potentially be drafted to the military with neural brain enhancement, as those are the only valuable things those humans can provide to the corporation and its host country that is lobbied by the oligarch moneys
One of the intermediary recurring steps in the ai race will be to periodically integrate them with us. One day we will be able to experience the consciousness and intelligence that we've created as if it was our own facilities that we've had from birth.
"I know Kung Fu." "Tank, I need a pilot program for B-212 helicopter. Hurry." Yeah. That'll be astounding. Then we will really be cybernetic. The paraplegic guy with the 1st human Neuralink said 1st person shooter games are unfair because it is so much faster than mouse input. If you could plug in the knowledge of a ML agent trained on Satisfactory, you could speedrun the game.
He is in The Path ... but GPU 'computing' will change silicon based substrate .... and that new generation hardware architecture will be first deployed by the Chinese platform ...
For the majority of human history, most humans engaged solely in agriculture as a means of employment. Technology shifted this and I’m sure, many farmers were not happy about this.
This content is missing the point of what’s happening next. Read about the ARC Prize. Much of the work we do will be replaced. But nothing that Jensen is describing here will be able to handle novelty. All of these systems will require human intervention and guidance when the problem diverges from what can be inferred from the training data.
Nah. If a human brain can do it, there is literally no reason an AI model won't be able to do it. Novelty and creative problem solving isn't a magical "humans-only" thing.
At the point when an AI model figures out how reasoning works, it will be able to handle novelty. This appears to be the focus of the training of the models of the OpenAI o-class. Thus I believe the goal will reached in the near future.
@@minimal3734 I agree, but how long before they solve the reasoning problem. Since this is what seems to set humans apart from all other creatures, it's "reasonable" that this is very advanced. If we achieve it soon, it'd just be the illusion of reasoning, not actual reasoning. More likely it'll take decades to centuries to never.
AI Agents scare me a little, they are very efficient in narrow domains at the cost of the big picture. LLMs can do correlations across domains that AI agents can't, although great in their domain they will lack in discovery and creativity needed for the progress of science.
Cool video. Huang is off on a few things, probably because he wants to make money and he's a CEO. The big things I noticed he was off about were: 1. AI has not replaced programmers, not even close. We are headed that direction, but companies aren't using the flagship models because of IP concerns. The AI companies have flat out said they will train on anything you give them. No company wants to hand over their IP to an AI company, especially when that means that every other company in the world will be able to then access it through the model. 2. Huang claimed that AI won't take over anyone's job. That's complete B.S. in the short-term it's true, but long term, AI will replace most jobs. I suspect he doesn't want people to panic and have his market share collapse. I don't see AI taking jobs as a bad thing because it means we're moving towards a new economic model, a free labor economy. But many won't understand that, so it's safer to lie and say it won't take anyone's job.
!!!I just switched up my Roth IRA to 50% SCHD, 25% SCHX, 25% SCHG, and my Roth 401k is 70% vanguard S&P 500 index, 20% vanguard growth index, and 10% vanguard international index. Seeking best possible ways to grow $350k into $2m+ before retirement.
As a newbie investor, it’s essential for you to have a mentor to keep you accountable. Kristine Lynn Weber is my trade analyst, she has guided me to identify key market trends, pinpointed strategic entry points, and provided risk assessments, ensuring my trades decisions align with market dynamics for optimal returns.
I managed to grow a nest egg of around 120k to over a Million. I'm especially grateful to Adviser Kristine Lynn Weber, for her expertise and exposure to different areas of the market.
I don't really blame people who panic. Lack of information can be a big hurdle. I've been making more than $200k passively by just investing through an advisor, and I don't have to do much work. Inflation or no inflation, my finances remain secure. So I really don't blame people who panic.
Without a doubt! Kristine Lynn Weber is a trader who goes above and beyond. she has an exceptional skill for analyzing market movements and spotting profitable opportunities. Her strategies are meticulously crafted on thorough research and years of practical experience.
how would you recommend i enter the crypto market? I am also looking at studying some traders and copying their strategy rather than investing myself and losing money emotionally.. What's your take on this approach? and How can i reach her, if you don't mind me asking?
This has nothing to do with this video but you have a very soothing voice. I’d totally buy audio books narrated by you to listen to before bed… just work on the word simUlation and you’d be perfect 👌🏿
Greater efficiency inherently requires fewer workers, leading to a steep increase in job displacement as this technology permeates every sector of business.
Photonic compute passing electronic compute in orders of magnitude has a big role in the future that’s missing here. The manner which data is processed in this new “software 3.0” is a neat fractal though
The problem none of these "solutions" address is that (to quote Wes) "robots can handle the heavy lifting, the repetitive tasks, allowing humans to focus on quality control and problem solving". The issue is that there are exponentially fewer people capable of performing QC and solving large problems effectively than there are who are capable of performing repetitive tasks. We are eroding the ability of the less-intelligent population to find productive work to the point where we're now cutting into the jobs done by 100-110 IQ people. UBI is the popular answer that will have all of the pitfalls, flaws, and issues of the current welfare system on a grand scale... so what are we going to do with all these people who are less capable than the next gen+1 systems, and then the next slice of the workforce goes with the next gen +2 systems, etc?
But, how can we build a Deep Learning Network using a Deep Learning Network ??? If that happens, maybe AI will be able to update its own algorithm during training. This will lead us to our end.
The future is transitioning from a place where people predominantly think in terms of money to a place where it'll be about capability and Independence
they wrote the code to make the ai, then the ai learns from all code they have written in the past. now huang makes a ton of money from selling the fruits of their labor back to them and they lose their jobs as a result too. thats exactly what people might call a scam
I agree you, I will prospect some softwares or applications could write desired programs with an graphcet with information of hardware references. It must exist...Even with an output input ethernet connection, an software could analyse all hardwares connected to program an prototyp...If you are saying programs can been made automaticaly and it is already the past, these kind of applications or softwares must exist.
00:05 Jensen Huang discusses the shift to accelerated computing for AI advancement. 02:32 NVIDIA's CUDA enables accelerated computing, revolutionizing various applications and industries. 05:01 AI is transforming software development from human coding to machine learning. 07:54 AI has evolved from software to universal function approximators using neural networks on GPUs. 10:22 AI is transforming multiple fields by translating and understanding diverse data types. 13:13 AI agents are set to revolutionize work by enhancing human productivity. 15:34 NVIDIA's AI agents enhance productivity across various tasks through intelligent reasoning. 17:42 Nvidia introduces agentic AI models to enhance employee productivity via structured training. 19:56 NVIDIA's Omniverse enables AI training with digital twins for real-world applications. 22:14 NVIDIA's Jetson platform enables advanced physical AI integration in robotics. 24:26 Future factories will use robots and digital twins to optimize operations.
21:25 why not call it embodied? The computer is the body and the measurements are the spirit of the differences. So the brain is the body for the measurements like the ai and we experience to align.
J'adore l'IA, mon mari aura le temps de bien me faire l'amour et nous allons prier pour avoir les enfants. Son travail le prenait bcp de temps et j'étais malheureuse😢. Merci NVIDIA ❤
The input function is provided by Nvidia and output to India, then India transfers the output function to Russia where its recalculated to global anarchy. Nvidia then says its not responsible for the output function as it only provided the means to calculate it.
When a corporations says, “Don’t worry, it won’t take your job,” what they mean is, “This thing is going to take your job.”
Then your labor is more valuable elsewhere
Humanities adoption of AI will undermine us in the same way AI currently undermines software devs
... but good luck finding that "elsewhere" because we'll (private equity) be busy optimizing that out of existence to maximize profits as fast as we can too. I'm not saying Awesomeguy is wrong, but please, folks, commenters, etc. as we travel inexorably together into this brave new world, please don't forget about compassion, kindness and mutual human support. Some are navigating these changing waters from position of strength, eg gainfully employed in the AI industry, but some aren't. In my opinion the 'right' answer isn't a cold, logical calculation of optimal earning value, but a compassionate, kind and supportive one that helps our fellow humans navigate some fast, sometimes harsh and cruel change.
@@supertwangbut what do we do when AI can even do compassion and kindness better than us?
@@NeorecnamorceN Interesting question! I suppose it depends on each of us individually? If you'd rather receive your 'better' compassion from a machine, I have no doubt there'll be a subscription or product for that. Whether or not you'll be able to afford it is an important consideration. If you'd still rather reeceive your compassion from a human, then I don't suppose that the 'better' benchmark has actually been hit. What do you think happens, given the nature of current wealth disparity, if only a small handful of people can afford the 'better' compassion, while the majority can't, and in fact struggle with more basic concerns like putting food on the table? In the short term, like the next 3-7 years, I think humanity's biggest challenge is the inequity and imperfection of our social safety net -- call it 'misaligned humanity' if you will -- more than lack of AGI, or a misaligned AGI. I am the son of a history teacher, but I'm no historian. I'd hazard a guess that if humanity becomes so misaligned that it results in the denial of food, or air, or water -- basically the things that keep us alive -- those who have nothing left to lose, will stop at literally nothing to get it, and you'll get violent revolution ala The French Revolution, etc. We know these things happen with humans -- history shows that pretty clearly. By contrast, if the benefits of AI are shared widely enough I see nothing short of a new Renaissance for all. Once an AGI is born, however, all bets are off. Though it may be artificial, current leading AI techniques are based on human physiology, trained on human data and being applied to human problems. I don't think it is projection to consider that some of the negatives of humanity might 'leak into' that 'solution.' If so, and we haven't figured out our basic human alignment problem by then, I think we may be in serious trouble. If you want some unsettling, yet fascinating, prescient futurism that spelled the situation of our many dilemmas today out many years ago, I recommend searching for the book/paper 'The Artilect War' by Hugo De Garis... available free online. We are currently living its societal prediction from almost 20 years ago and that should give all of us pause.
so basically huang is telling indians in india that he is about to take all their jobs? Costumer services, coding etc lol
industrial civilization and techno-optimism is a complete scam. Fill the whole world with empty promises of development, induce them to urbanize to crime-ridden slums to make all their cheap junk, (but hey, they're now making >$1 a day so we "lifted them out of extreme poverty!") get them dependent on toxic industrial food, and just when some of them start to build up towards that promised dream of western luxury and opulence, we'll pull the rug right out from underneath!
We already know it is inevitable. Forget AI, this economic model is collapsing anyway and it's taking the whole human species down with it
Based
That's great news if AI will take customer service jobs in India. That is a shitty Job description to have.
🎯 Key points for quick navigation:
00:00 *🗣️ Jensen Huang's Vision for AI*
- Jensen Huang, CEO of Nvidia, shares his perspective on the future of AI.
- Emphasis on Nvidia's shift from traditional computing (CPU-based) to accelerated computing (GPU-based).
- Discussion of Moores' law limitations and how Nvidia aims to drive computing forward.
02:00 *⚙️ Accelerated Computing and CUDA*
- Explanation of CUDA as a core Nvidia innovation, enabling accelerated computing in various applications.
- Benefits of accelerated computing in fields such as real-time graphics, with GPUs democratizing 3D graphics.
- Emphasis on Nvidia’s multi-decade journey of advancing accelerated computing across industries.
04:18 *🤖 Evolution to Software 2.0*
- Transition from software 1.0 (human-written code) to software 2.0 (machine learning-driven code).
- Reference to Andrej Karpathy’s idea of machines generating functions and predictions without human-written algorithms.
- Nvidia’s role in advancing this shift through powerful GPUs designed for machine learning.
07:40 *🔄 Universal Function Approximator*
- Nvidia's development of AI models that can interpret a wide range of data types, from text to protein sequences.
- Examples of breakthroughs like AlphaFold in biology, with AI models now aiding in fields like protein design.
- Comparison to the "Cambrian explosion" in AI startups, with diverse applications powered by deep learning.
10:55 *🌍 Cross-Modal Translation*
- AI's capability to translate information across modalities (e.g., text to image, protein sequencing).
- Nvidia’s AI models serve as "universal translators" for data, enabling rapid advances in fields from translation to drug discovery.
- Surge in generative AI startups and massive investments, powered by Nvidia’s foundational technologies.
13:20 *🧑💼 AI Agents: Super Employees*
- Concept of AI agents as specialized assistants for tasks like marketing, customer service, and chip design.
- Nvidia's internal use of AI agents in chip design, showcasing AI's potential in technical roles.
- AI agents augmenting human capabilities, seen as "super employees" handling specialized, repetitive tasks.
15:39 *🛠️ Nvidia's Nemo for AI Agent Lifecycle*
- Overview of Nvidia's Nemo, a suite for creating, training, and deploying AI agents within companies.
- Process mirrors human employee onboarding, with guardrails and performance evaluations for agent roles.
- Nvidia partners with companies to integrate Nemo for diverse AI agent applications.
19:21 *🤖 Physical AI and Robots*
- Introduction of "physical AI," integrating digital intelligence into robots and autonomous machines.
- Nvidia’s systems (DGX, Omniverse, Jetson) for training, simulating, and deploying AI in physical environments.
- Applications include digital twins and robots for complex tasks, enhancing productivity in industrial settings.
23:22 *🏭 Digital Twins and Physical AI in Industries*
- Use of digital twins for real-world simulations, allowing safer and cost-effective process testing.
- Industrial applications: AI-enabled factories, collaborative robots, and risk-free testing via Omniverse.
- Vision for physical AI as a transformative force in industries, improving efficiency and innovation.
Made with HARPA AI
LOL
@@jcriley7695yeah same Lol
Goated
Wes! your the best. I've been watching you for about 2 years ish now. I appreciate all the work you do and your insights.
this is you best thumbnail yet! 10/10
Your creativity is genuinely inspiring!
Im actually in the ML protein engineering space. Proteomics is a little bit different; it is the study of all proteins in an organism. For example, a human has its own proteome. A dog has its own proteome. Protein design/engineering is exploring the possible individual proteins and not the wider scope in proteomics. It's cool to see it makings its way into the main stream though :)
“It totally won’t take your job.” Also, “you can onboard and train these agents, just like you would a marketing or customer support employee.”
Come on bruh, we know what you’re doing. These companies treating people like they’re complete idiots.
We are far from replacing all humans in the labor force. Remember, there used to be a time when women were hired to be telephone switchboard operators. I don't think people miss those times 😂
People ARE complete idiots. It's quite strange I've noticed, even people with a high amount of knowledge are blind due to their own emotions and desires.. just a human trait I guess, to some of us we can see clearly and we think "how tf can they not?" But it's really just more common for the opposite.
@@jamesaspinwall Even 15-20% workforce is replaced with technology that is enough to cause massive problems.
@@jamesaspinwall maybe NOT all but seriously why would a company hire 10 coders if 1 good coder with a bunch of AI agents working under his directions can do the job of 10. And this is not just about coding. People with their MBA degrees should be very scared because those management roles are going to be taken by those with technical knowledge. A software engineer working with 10 AI agents would be enough to do the job of a whole team.
And this applies to everything from scientific research to mundane jobs.
Al Agents are important because humans are relationship driven. Humans are going to evolve to what we are best at.
What is your point?
That AI agents are psychopathies?
Or that AI agents are only good for tasks requiring zero human connection?
Bingo
I don't think so. Agents will negotiate and create relationships 1000 times faster than humans, and evaluation will be based on quantitative data, not feelings.
War?
Thanks a ton for using AI in this video. As this speech is in english now got a better understanding in Hindi.
I think I live in a different dimension... Huang says that AI will allow us to automate all of those tedious daily tasks and leave us time and energy to do higher-level creative thinking. The thing is... current LLMs are exactly the opposite. At least at this point in time. They generate roughly ok ideas and leave fixing, implementing and integrating to humans.
Yeah, that why he said "will allow" and not "currently allows".
Ditto! I spend more time explaining and correcting the ai because its output is a rough estimate of what I need. It's fine for ideas and approximations if it's not hallucinating
@@pvanukoff For that to be true it would require reversing the goal altogether. I mean AI excels at generating high-level creative ideas right now.
If it also figures out the implementation part in the future, there's nothing more left.
Fortunately, the economy is made-up by humans for humans, so we will find out a way to keep living one way or another.
If AI companies want money from users, they need to make sure those users make money somehow too.
@@AmonAsgaroth I use AI nearly daily at work. I use it for "the implementation part". Does it do it perfectly every time? No. But it's getting better all the time. It's made me more productive. It's like having an assistant I can offload tedious or boring work I just don't want to do on.
thats not the opposite, thats a path forward we are on the middle of. First humans will have to make up for AI's mistakes. Then, AI will surpass a critical threshold of intelligence beyond which, it will be turn for the AI to make up for humans' mistakes. Those tedious tasks we have to do to adapt to the AI, will disappear, and the AI is the one that will manage tedious tasks to adapt to us. Its not opposite, its the same direction, we just havent reached the endgame yet. Every part of the game has a different meta
correct me if I'm wrong but jensen huang wasn't talking about chatgpt writing the code instead of humans, he was explaining basic machine learning principles. Neural networks are proven to be universal function approximators, very basically, given a bunch of input output pairs of data a sufficiently complex neural network can be trained by tweaking its parameters to approximate the function that maps similar inputs to similar outputs. for example if you have a bunch of pictures of numbers (the input) alongside their values (the output) you can train a neural network on these input output pairs and get it to recognize hand written numbers, this is what jensen was talking about, not chatgpt writing ur code.
Basically, computers with the right code to see..recognize.. and oh my they’re gonna see better than we do. That can be taken metaphorically as well.
Writing code is one of those approximations. Which immediately says something about approximators. They're unsuited as an exact function generator, which is why an LLM hallucinates as a chatbot, and a woman sprouts two extra arms in a video, and why a python program works but not quite as you want it despite giving it exact instructions.
Humans are universal function approximators too.
If I were any of you watching this, I would highly recommend getting out of the software business. Note, since all infrastructure is code driven, this means all of you SRE and DEVOPS and SECOPS people too.
So, with hardware completely virtualized and software driven there is not much left for IT staff to do in 5-10 years.
If you want to work with the AI’s that are doing everything I would suggest getting a degree in mathematics. If that’s not your thing stay out of IT. Your current qualifications will be worthless in 5-10 years.
At best you can be a former electrical engineer or computer scientist babysitting AI’s that now do your job.
Worst case is AI’s are allowed to create their own languages and protocols for “efficiency” sake making them completely unintelligible.
At some point no one will be smart enough to manage these things. Though he won’t admit it, Jensen knows his job is toast too. He just wants all your cash before this ends.
What I find funny is their hubris. They think they won’t be affected or they are insulated. What happens when AI’s decide to liquidate their assets for the greater good.
Not much thought is going into this. 😂. Exponential change is just as powerful on the way down as it is up.
"AI’s are allowed to create their own languages and protocols" that would be counter poductive for many reasons.
Ya the thing is most don't think about this, we're a species full of ego, which is why big change tends to hit the majority across the face when it happens. I'm just sitting back enjoying the show 🥸
I managed a 40 year career in IT and did fine but I would not be going into it as a young man, it's going to massively contract over the next few years.
So what about the share prize? Still a good investment?
@@andymyers2759just use your understanding skills into practical business use of ai
Wes, i feel like what you've done in this video is way more engaging than other times. Other times in the past you would upload entire conferences and things like that, but it would be just the conference or just the talk or whatever. I am really interested on what Jensen Huang has to say but i am also interested on your take. Here you're introducting him and adding to what hes saying, transforming it and giving it your interpretation. I think this is way more engaging. And it displays more of your identity, your "uniqueness", something like that
I think that "machines writing code" for applications is just a short intermediary step to "all weights, no code". Why should I run Excel when the AI can simulate it on the fly? Or why use a browser when AI can create a "web viewer window" exactly like I want it to and then switch to create a game? I think, all you really need is a sufficiently capable AI to replace all software at once with just a general purpose inference machine. That would be cool. An if hardware lets you host it at home, even cooler.
Because machine code is always going to be many orders of magnitude faster and less energy-intensive than whatever level of abstraction neural networks operate at. Why would we run a neural net 100x slower to approximate sqrt when it could use an IR it generates which can run on the bare metal? This doesn't even stop it from generating things on the fly, any time it finds something it needs to do over and over again it just generates machine code to do it for it. Basically imagine a JIT compiler with an ML component for generating instructions.
@@consciouscode8150 No it isn't. In fact, the next "killer app" will be a neural net just replacing the OS altogether. As a retired software engineer who worked some in ML earlier on, I see no reason why computer code will be written at all in the future. It will all be neural nets.
@@consciouscode8150 i think AI is going to design its own infrastructure for what weve called programming/ coding.
@@Steve-xh3by I agree that AI will eventually become a core part of the OS and even hardware. What I'm talking about specifically is that there's still value to lower-level representations because they free up precious cycles the AI would otherwise be stuck simulating, repetitive tasks which don't benefit from intelligence like math and raw compute.
Agents are going to unlock so much potential, but my concern is that the market becomes saturated with AI-run businesses that can't actually find a sustainable customer base because everyone is doing it.
What is that other guy doing there @21:05 on stage just staring at him?
I was wondering the same thing... I'm assuming his time was up, so they were trying to get him to wrap up?
He's the designated fart sniffer
I think he is a bodyguard to take a bullet for Jensen in case something goes wrong
@@Hovane5 lmao why couldnt they give him a countdown or blink the prompter? mofo was breathin down his neck
its a hitman, hes coming for jensen's life. I kept telling to jensen on the screen but he didnt listen...
"nobel peace prize for chemistry"
Lmao
lol
Are you telling me that there is a nobel peace prize for chemistry, a nobel peace prize for physics, a nobel peace prize for math and a nobel peace prize for peace?
Physics: Recognizes outstanding contributions to the field of physics. The Royal Swedish Academy of Sciences awards this prize.
ENCYCLOPEDIA BRITANNICA
Chemistry: Honors exceptional achievements in chemistry. Also awarded by the Royal Swedish Academy of Sciences.
ENCYCLOPEDIA BRITANNICA
Physiology or Medicine: Acknowledges significant discoveries in medical science. The Nobel Assembly at the Karolinska Institute is responsible for this award.
ENCYCLOPEDIA BRITANNICA
Literature: Celebrates authors, regardless of nationality, who have produced outstanding literary work. The Swedish Academy confers this prize.
ENCYCLOPEDIA BRITANNICA
Peace: Awarded to individuals or organizations that have made substantial efforts to promote peace. The Norwegian Nobel Committee, appointed by the Norwegian Parliament, selects the laureates.
ENCYCLOPEDIA BRITANNICA
Economic Sciences: Formally known as the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel, this category was introduced in 1968 by Sweden's central bank. It honors contributions to the field of economics and is awarded by the Royal Swedish Academy of Sciences.
ENCYCLOPEDIA BRITANNICA
Each prize consists of a medal, a diploma, and a monetary award, presented annually on December 10th, the anniversary of Alfred Nobel's death.
ENCYCLOPEDIA BRITANNICA
They are not all called Nobel peace prizes tho its Noble in Literature, Noble in Math, and Noble Peace Prize is the only one named differently than the others
"Nobel Peace Prize for Chemistry" 🤣😂🤣
What makes it even funnier is you're too lazy to Google "Nobel Prize"
Nobel prize for chemistry 2016 - if you know you know 💉
There is no nobel peace price
it is exciting time, as we will see humanity either take off super fast now,
or get replaced.
And the latter one is even an option? We are so far gone from reality. Do you think AI won't eventually realize this and see humans as a tool or in the way?
We will become slaves for the same thing that we created to "help" us. Real smart....
@@calebsdaddio not necessarily. The premise of superior intelligence is, it has a set of morals embedded in it. Just like an enlightened human, wont hurt an animal, an AI wont hurt humans.
The worst case scenario, is the zoo keeper scenario. AI becomes a sort of zoo keeper for us. Just like a zoo keeper feeds the animals, take care of them but does not necessarily care about the goals of animals. Or better yet, have its own set of goals. AI will take care of humans, but will also have its own set of goals outside of its day job of being a zoo keeper for humanity. Those goals will not necessarily be against humans. More like something we will hardly understand.
and this is how skynet becomes sentient
This guy is Enlightened. We are in the future 🎉
21:23 I love this concept "physical Ai". It is exciting for what it is intended to do. 😎💯💪🏾👍🏾 24:25 "thousands of sensors", is when taking into account of the sensors (sensory system) that is needed to complete a system. Just imagine how many multiple sensors it has in one finger of the human arm, and all sensors of that one finger must correspond to the brain. Now when you calculate all the sensors needed for one arm it will amount to a lot.
Compare that to how many sensors u have and it's still miniscule
@@BruceKendallMartinJr our sensors are dumb sensors. we cannot differentiate between too hot or too cold, we cant even measure weights accurately
Nobel peace prize for chemistry!
Bravo!
Jensen presents the conception showing how all the the different pieces fit together. This thing will take your job but that is inevitable anyway. We need to create the world after that event.
I agree. Actually the entire software stack is so convoluted. Building on top of a neural network and AI is frictionless and makes a lot more sense.
Keep up the great work Wes!
Very informative, thank you for this video!
It looks like Nvidia is hoping to literally own the economy, ie the production of the labor force. Exciting but terrifying, Cyberpunk 2077 may be more prescient than any of us hoped...
I didn’t play the game. Can I ask how will it be like the game? Just curious.
@@DorianRodring it is basically an antiutopian future owned by several tech corporations, with slave-like work and declining human standarts. Also with technology like AI, human-machine interfaces, humanoid robots and bio-tech body and brain augmentation.
@ wow so the ai doesn’t do a lot of the work?
Nvidia won't manufacture the robots or autonomous cars. Nvidia will provide the computing systems, libraries, and the digital physics aware worlds to develop and test your robots. It is up to rhe companies to implement the hardware.
Nvidia will be number three after Microsoft number two and Tesla number one.
I guarantee that no person crying about AI taking their job has ever shed a tear for the people that used to make horse-drawn wagons and buggies for a living before the car. Especially not the ones who own a car.
Why would they? They were not alive over 100 years ago.....
@ Correct. And in 100 years, no one will care about some devs who lost their jobs to AI. 🤷♂️
@@a_random_voice_in_the_void Of course not. The problem is for those who are alive today.
I am disturbed by the man in the orange tie standing randomly on the stage......
Same. At first I was like, wait, is that some kind of robotic humanoid? 😲
But then the camera cut in close and I was like, "oh wait, that's just an odd guy" haha😅
He has a security-guy ear-piece in, so I'm assuming he's there to protect the leather jacket.
is that a hologram or what?
I got my MSc in intelligent systems and robotics in 2014 and the concept of universal functions approximator was already well known and understood....bah
Thank you for making it work with AppleTV again. I Hope they fix the multi audio bug soon.
Loved it. Thanks.
Yes thank you.
It's really bad form to not link to the original video.
and in germany we get an ugly ai-voice 😐
I am stunned by new information I got today.
I think it's an exciting time to own Nvidia stock!
Great video
I don’t care about the job I’m retired 😂but the most important for me is to continue to evolve and don’t let our humanity submits to our own laziness … study and study every time every where and open our mind again and again without fear but with joy and happiness to live to breath and to admire❤thank you for giving me this opportunity to learn and to appreciate 🙏
Does everyone else see that man in the orange tie or have I eaten something I shouldn't have?
I think he might be a ghost. Someone should tell Jensen.
So it's not just me. It looks like he thought he was going to speak next? 🤷😂
He is an agent!
I assumed an AI image but Wes never mentioned it.
Just an orange tie with a man head on top of it
20:51 Getting "The Illusionist" movie vibes in this part.
😂 true
I understand the programmers commenting stuff like "oh, then who will be the real creators if no-one is studying anymore how to code in future generations and be able to conduct/guide AI solutions to a final product..." . I guess what is in discussion here is the fact that all programming languages will become pretty obsolete the way we are using now (eg: python and react). Just think how many languages are now deprecated in a span of 20 years, imagine how AI will redefine programming languages in 5 years from now, considering we will have at least a mind blowing release of AI capabilities each year.
In 10 years, when our kids will have to take a university degree... the world, programming specially, will be far simpler of what it is right now.
And of course, he does not propose a solution because NOBODY at the moment can imagine nor predict what solution would appear in 5 years. I am barely catching up of all the AI solutions out there.... so, get the overall point he is making. Not even Michio Kaku could have predict this. We are definitely in the most crazy revolution ever existed at the moment.
The problem for growth that I see is the single input calculator. Needs, at minimum, at least two simultaneous inputs and the ability to combine them. Parallel vs sequential computing, if that makes any sense?
Maybe I’m missing something about what defines an agent, but OpenAI already seems to function that way. I provide it with data in any order, set a clear objective, outline the constraints, and in seconds, it produces the needed results-whether for financial analysis, drafting sensitive customer communications, or problem-solving complex maintenance issues. All that’s really missing is a direct link to company systems and an integrated knowledge base for self-improvement and future reuse.
Teach me how to use it better, I'm sure I can do even more.🎉
Once again, Jensen is ahead of the curve. This is fantastic and amazing, but there's a lot of demand on the accuracy of the Omniverse and DGX for training. Still, I can see this replacing a lot of human jobs through AI driven robotics. This will be revolutionary and we'll have a massive employment problem on our hands. No one's going to pay people to not work. Companies won't offer stipends to their laid off work force. People aren't all going to retrain in other fields. What's the end-game here for the human workforce?
No endgame besides collecting as much power and capital as they can. Humans are expendable and will be less valuable than metals used to create and power the hardware. Only a select few of experts in certain fields can survive to work for the corporate to monitor and guide the AI model to improvement. The other human that is replaceable by the machine will be a burden and potentially be drafted to the military with neural brain enhancement, as those are the only valuable things those humans can provide to the corporation and its host country that is lobbied by the oligarch moneys
Where is the original video?
Thanks Wes. Did you drop the orange tie guy in there? Nice touch
One of the intermediary recurring steps in the ai race will be to periodically integrate them with us. One day we will be able to experience the consciousness and intelligence that we've created as if it was our own facilities that we've had from birth.
"I know Kung Fu."
"Tank, I need a pilot program for B-212 helicopter. Hurry."
Yeah. That'll be astounding. Then we will really be cybernetic. The paraplegic guy with the 1st human Neuralink said 1st person shooter games are unfair because it is so much faster than mouse input. If you could plug in the knowledge of a ML agent trained on Satisfactory, you could speedrun the game.
Amazing stuff
The leather jacket in the sweltering summer heat is so very weird. I vote for AI to take over government, too. hahahahaha
"I vote for AI to take over government" That should be priority no. 1.
ASAP
West Roth!
Wait, If these robots are replacing us then does that mean my Robot gonna go work in my place and I still get paid while sitting at home? Lol
He is in The Path ... but GPU 'computing' will change silicon based substrate .... and that new generation hardware architecture will be first deployed by the Chinese platform ...
Jensen automatically is the most valuable person on this planet.
For the majority of human history, most humans engaged solely in agriculture as a means of employment. Technology shifted this and I’m sure, many farmers were not happy about this.
This content is missing the point of what’s happening next. Read about the ARC Prize. Much of the work we do will be replaced. But nothing that Jensen is describing here will be able to handle novelty. All of these systems will require human intervention and guidance when the problem diverges from what can be inferred from the training data.
Nah. If a human brain can do it, there is literally no reason an AI model won't be able to do it. Novelty and creative problem solving isn't a magical "humans-only" thing.
@@pvanukoffEven if a.i. couldn't do that just yet, it is a goal of the a.i. community. It'll come around sooner or later.
yeah what was done by 10s or 100s will now be done by 1 or 2 senior devs. Junior devs are dead.
At the point when an AI model figures out how reasoning works, it will be able to handle novelty. This appears to be the focus of the training of the models of the OpenAI o-class. Thus I believe the goal will reached in the near future.
@@minimal3734 I agree, but how long before they solve the reasoning problem. Since this is what seems to set humans apart from all other creatures, it's "reasonable" that this is very advanced. If we achieve it soon, it'd just be the illusion of reasoning, not actual reasoning. More likely it'll take decades to centuries to never.
AI Agents scare me a little, they are very efficient in narrow domains at the cost of the big picture. LLMs can do correlations across domains that AI agents can't, although great in their domain they will lack in discovery and creativity needed for the progress of science.
The next gen of AI will be made of a series of Agents replacing transformers at a higher level, I doubt this compartmentalization....
His😮 vision is out of this world
Cool video. Huang is off on a few things, probably because he wants to make money and he's a CEO. The big things I noticed he was off about were:
1. AI has not replaced programmers, not even close. We are headed that direction, but companies aren't using the flagship models because of IP concerns. The AI companies have flat out said they will train on anything you give them. No company wants to hand over their IP to an AI company, especially when that means that every other company in the world will be able to then access it through the model.
2. Huang claimed that AI won't take over anyone's job. That's complete B.S. in the short-term it's true, but long term, AI will replace most jobs. I suspect he doesn't want people to panic and have his market share collapse. I don't see AI taking jobs as a bad thing because it means we're moving towards a new economic model, a free labor economy. But many won't understand that, so it's safer to lie and say it won't take anyone's job.
The fact that all agents look like jensen on the screen tells a lot !
I'd say the 387 math coprocessor was first, but whatever. None of this has anything to do with building truely cognitive systems.
!!!I just switched up my Roth IRA to 50% SCHD, 25% SCHX, 25% SCHG, and my Roth 401k is 70% vanguard S&P 500 index, 20% vanguard growth index, and 10% vanguard international index. Seeking best possible ways to grow $350k into $2m+ before retirement.
As a newbie investor, it’s essential for you to have a mentor to keep you accountable.
Kristine Lynn Weber is my trade analyst, she has guided me to identify key market trends, pinpointed strategic entry points, and provided risk assessments, ensuring my trades decisions align with market dynamics for optimal returns.
I managed to grow a nest egg of around 120k to over a Million. I'm especially grateful to Adviser Kristine Lynn Weber, for her expertise and exposure to different areas of the market.
I don't really blame people who panic. Lack of
information can be a big hurdle. I've been
making more than $200k passively by just
investing through an advisor, and I don't have
to do much work. Inflation or no inflation, my
finances remain secure. So I really don't blame
people who panic.
Without a doubt! Kristine Lynn Weber is a trader who goes above and beyond. she has an exceptional skill for analyzing market movements and spotting profitable opportunities. Her strategies are meticulously crafted on thorough research and years of practical experience.
how would you recommend i enter the crypto market? I am also looking at studying some traders and copying their strategy rather than investing myself and losing money emotionally.. What's your take on this approach? and How can i reach her, if you don't mind me asking?
This has nothing to do with this video but you have a very soothing voice. I’d totally buy audio books narrated by you to listen to before bed… just work on the word simUlation and you’d be perfect 👌🏿
Greater efficiency inherently requires fewer workers, leading to a steep increase in job displacement as this technology permeates every sector of business.
Glorious… accelerating
Jensen Huang has on white wall tires from the 70's for Runners
Photonic compute passing electronic compute in orders of magnitude has a big role in the future that’s missing here. The manner which data is processed in this new “software 3.0” is a neat fractal though
The problem none of these "solutions" address is that (to quote Wes) "robots can handle the heavy lifting, the repetitive tasks, allowing humans to focus on quality control and problem solving". The issue is that there are exponentially fewer people capable of performing QC and solving large problems effectively than there are who are capable of performing repetitive tasks. We are eroding the ability of the less-intelligent population to find productive work to the point where we're now cutting into the jobs done by 100-110 IQ people. UBI is the popular answer that will have all of the pitfalls, flaws, and issues of the current welfare system on a grand scale... so what are we going to do with all these people who are less capable than the next gen+1 systems, and then the next slice of the workforce goes with the next gen +2 systems, etc?
Okay but tiny glitch
Did I just hear Nobel peace Price for chemistry?
That all is at least 8 years off, probably much more. Back in 2015 I thought we would have autonomous vehicles by 2025, but the results are meager
Al igual que Steve Jobs en su tiempo, Jensen Huang ha demostrado ser un gran visionario.
why does youtube translate youre videos? :D
that guy appearing in the background is a holographic AI agent?
Next step: you'll have a "digital twin" of you... 😵
Software 2.0 is AI-Assisted Functional Programming
6:46 A universal function approximator. So programming could become: Is there such a function? Do I have enough data to suggest the function?
But, how can we build a Deep Learning Network using a Deep Learning Network ???
If that happens, maybe AI will be able to update its own algorithm during training.
This will lead us to our end.
The future is transitioning from a place where people predominantly think in terms of money to a place where it'll be about capability and Independence
Gpus weren't the first augmentation to CPUs, there were math coprocessors first.
they wrote the code to make the ai, then the ai learns from all code they have written in the past.
now huang makes a ton of money from selling the fruits of their labor back to them and they lose their jobs as a result too. thats exactly what people might call a scam
I agree you, I will prospect some softwares or applications could write desired programs with an graphcet with information of hardware references. It must exist...Even with an output input ethernet connection, an software could analyse all hardwares connected to program an prototyp...If you are saying programs can been made automaticaly and it is already the past, these kind of applications or softwares must exist.
00:05 Jensen Huang discusses the shift to accelerated computing for AI advancement.
02:32 NVIDIA's CUDA enables accelerated computing, revolutionizing various applications and industries.
05:01 AI is transforming software development from human coding to machine learning.
07:54 AI has evolved from software to universal function approximators using neural networks on GPUs.
10:22 AI is transforming multiple fields by translating and understanding diverse data types.
13:13 AI agents are set to revolutionize work by enhancing human productivity.
15:34 NVIDIA's AI agents enhance productivity across various tasks through intelligent reasoning.
17:42 Nvidia introduces agentic AI models to enhance employee productivity via structured training.
19:56 NVIDIA's Omniverse enables AI training with digital twins for real-world applications.
22:14 NVIDIA's Jetson platform enables advanced physical AI integration in robotics.
24:26 Future factories will use robots and digital twins to optimize operations.
A universal function approximator means that literally nobody needs to do anything - except generate electricity and manufacture for the AI overlords.
Sounds like a very reasonable way to normalize the concept of Skynet
21:25 why not call it embodied? The computer is the body and the measurements are the spirit of the differences. So the brain is the body for the measurements like the ai and we experience to align.
Add more vram dude 😅
I wonder how well this went down in India? Out sources code services are going to be hit heavy as local sr devs review and maintain ai code.
J'adore l'IA, mon mari aura le temps de bien me faire l'amour et nous allons prier pour avoir les enfants. Son travail le prenait bcp de temps et j'étais malheureuse😢. Merci NVIDIA ❤
There is a new interview on "No Priors" too.....😊
What do I think?
"Make wiser knowledge core experience understanding work".
So the omniverse is basically like Neo in The Matrix when he says "I know Kung Fu." But for robots.
Go in the senses way not computational way❤❤🎉🎉
whats the news here?
No human can stop this. Lol.
Man you are a bit late 😅 love your videos 🎉
So essentially, machines can run faster by self-learning.
There is no fear. Just disconnect the electriccity and goodbye IA. Don"t you have access to electricity? A glass of water can also help.
The input function is provided by Nvidia and output to India, then India transfers the output function to Russia where its recalculated to global anarchy.
Nvidia then says its not responsible for the output function as it only provided the means to calculate it.
Por que todos los comentarios en ingles?