Very insightful video, I remember Andrej covering something similar in his video "Intro to LLM's". He referred to the new age of systems as LLM OS, but after this video, it seems he was referring to Software 3.0. Especially after seeing new frameworks like Agent Zero that allow for an agent to basically build out anything it needs, solve errors it encounters, It can delegate tasks to other agent instances, it has access to tools etc.
The entire concept of running a business with just yourself and AI agents are very interesting and seem lucrative but the thing I can’t wrap my mind around is - if everyone does this, its absolutely no way our world wide economy can withstand it. This will most definitely, imho, cause the emergence of a new type of economic model that will be a successor to Capitalism. Maybe a new economic model that is compartmentalized in that each market is gauged or valued based upon how their respective businesses help push, advance and evolve the market it is within, for example; A customer service compartment may raise or lower the value of a AI-Human hybrid company based upon how their practices help better or lower the quality of Customer Service. Things are about to change and I think for the better
Exactly. There is a finite amount of wealth to capture in the world. What happens when everybody has the intelligence and capability to start their own business? Nobody wins in the end, except the legacy big players who already have accumulated wealth.
Just let this blow your mind first. What do you wish for first for with a genie that has no limitations on wishes. You wish for infinite wishes. What do you do when you get your own ai driven robot. You obviously have it make more robots. Then you sell the robots that are made. Or lets try something a bit different. Lets say you tell your ai robot you need it to make a farm that would fit into a certain space (indoors or outdoors) give it specifics on caloric needs for your family monthly and have it build it and set it up as a closed system so you have most of your caloric intake accounted for year round. Now what happens when u excitedly tell your neighbors and they want it too. U lend em ur robot, and you now take out a neighborhoods worth of capital out of grocery stores, then what happens to the truckers, what happens to the farmers, what happens to all the downstream ripples caused by one action? New actions breed new economies, but we will be economies of scale, trading in real time commodity trading on local block chains. Imagine you specialize in tool making because you like the idea of working a forge, and instead you sell your tools on the block chain, then you take your "earnings" and then get food you need from your neighbor down the street who likes to farm. You get your entertainment locally, from the creative people all around you. People will always want to work when it is doing something they love. And this system will free you up to follow what it is you really want to do. A golden age is coming. Its not that far off, maybe 2~3 years 🤔
@@Mrv958Yup. If anyone wants to save their time, software 1.0 is a person writing a computer program for a specific task. Software 2.0 is a person transcending the need to program by training a specialized AI to do that same task more efficiently. Software 3.0 is a big, general, and open source AI (ie llama 3.1 large) that trains specialized AI for specific tasks. In other words, a distinction for distinction's sake.
check out the game Screeps, it's a game designed to teach Java Script, but because the community has documented everything for years, Chat GPT can just code it amazingly well right out of the box
This will be much bigger than any other business wave. The crazy thing is the foundations are here. People just don’t know how to use them. As a consultant and university educator, I tell people- the world has already moved your cheese, you just don’t know it! Scary for many. Exciting for some. Well done!
Not sure Software 3.0 will be big models training small models. I mean, yes. We will see a lot of that. But i don't see it as a step change up from 2.0, as 2.0 was a step change from 1.0. Karpathy himself warned about the diminishment of model effectiveness when overwriting, or abstracting from, model weights. If you want a better model, you retrain it with new data *and* old data. Get weights right for the whole enchilada. If you abstract a smaller model from a larger one, will there be a bias towards the mean? A lesser field of competence? I dunno. I've been thinking instead of 3.0 as self-improving, self-actualizing, self-optimizing models. Q-Star, perhaps. A little furthur down that path is AGI. I don't think this two-tiered model development idea is on that path. But maybe I'm wrong?
Whether it ought to be called Software 3.0 or something different, I am thinking along similar lines in that I think a key component of AGI will be the ability to learn on the fly (self-improvement), among the others you mention. "I've got to warn you, it's a learning robot. Every moment you spend fighting it only increases its knowledge of how to beat you."
Every piece of useful software you've ever interacted with is a result of careful planning and hard work of very smart problem solvers. What we currently call AI is just a fancy auto-complete engine. Just remember that.
Great video.❤ I work so hard to engage myself in someway of earning more income and stop depending on anyone. my family are happy once again and I can now afford anything for my family even with my Retirement. With a step of $50k and $17k weekly returns has been life changing, after so much struggles.
My outlook on money changed when I realized someone making $200K can retire broke & someone making $80K can retire a millionaire. With the current market movement.
Yea, I agree with the other posts, I love Wes’s videos but sometimes I get hung up on how long they are. Maybe he needs to hire an editor to help? I’d be curious what would happen if he took his own transcript as a rough draft and put it into ai to make it more succinct, then edit as necessary following that. I bet the results would be amazing.
Definitely not crazy; it all depends on the cost of running those product models. We simply can't have that trivia if the mean medium cost are to thousands of dollars a day! Thanks for the video! ❤
2:35 Hebbian learning (neurons that fire together, wire together) isn't much used for current _artificial_ intelligence. Instead, so-called backpropagation is the predominant algorithm for training AI. For an overview, see Figure 1 in "Backpropagation and the brain" by Geoffrey Hinton and coauthors.
My agent would simply scrape what I wanted off of Facebook, like just posts from people I follow. No ads, no nothing and very little info given to the zuck. Same for Google searches.
One of the hallmarks of mania is a rapid rise in complexity. AI has become a giant bubble waiting to pop. Most of this technology is overhyped in its utility, and the profitably of that which is useful is dubious, because the ways it’s leveraged are democratized and can be done for free, and the inferencing is highly competitive.
It's crazy good if you know how to use it, so the tech itself is not overhyped, I would dare to say it's even underhyped. What is overhyped though are some AI-based products like Devin. AI itself won''t pop, waiting for it to pop is like waiting for computer or phone or internet "bubble" to pop - not gonna happen.
@@Laszer271 The way some people are talking about AI now is the way that I remember some people talking about the Internet in the 90s - it's overblown, it will pop, any day now... Well, the tech bubble popped in 2000, but that was a miniscule bubble compared to where tech is now. And the Internet just kept exploding. Most people and organizations are just learning how to use/integrate AI. It's underhyped. The results in medical companies that have done so (with AlphaFold etc.), from two whole years of integration, have produced astounding results. So yeah, even the models we have now like GPT-4 or Claude Sonnet are enough to drastically change society over the next decade, and the models just keep getting better.
@@squamish4244 Completly agree. AI right now is sports car in the middle of the deep forest. People found it and are complaining that it can't do much and walking is much more efficient. They can't imagine that the infrastructure for the car to really take off is only just being built. GPT-5 is not needed, we don't really need faster car, all we need are roads.
@@Laszer271 Speaking of cars, don't even get me started on how some people think that AI and self-driving cars are a good analogy, and "Look at self-driving cars, we don't have them yet, therefore AI is overhyped." There are so many problems with that analogy, good lord! GPT-5 is coming anyway, and LLMs went from "amazing" to "meh" in AI popular discourse in less than two years, but they're already moving past LLMs (LLMs only), so we have A LOT of building out of infrastructure to do in the next decade.
Love your philosophical takes! Do you think it will be more procedural generated? Software does not have to go away.. We need a workbench (comadore amiga reverence) for certain tasks right.. A photoshop 3.0 (or 2.0 for we are still stuck in a program from 1990) we will give a command and the really small procedural agents do this tasks.. We will guide them.. Little bit to the left or right give guidance, but then it will be more like clay.. But the workbench still needed right? Not all is abstract thinking in words... Love your work... respect!
Yep, I'll keep repeating it over and over. We can have a manageable, Zero-Loss AI transition through both sensible regulation and by the fact that AI is going to enable MORE jobs than it would have replaced. We have years of work converting not just hearts n minds to AI use, but businesses themselves. Use cases to be discovered, data hoards unleashed from private companies. I know of two geologists who are building AI based on centuries of company mining data, and they're gonna be able to send teams out and pinpoint sample pattern areas. Success means more jobs especially if the company hits paydirt at multiple sites.
To consider: programming language is a intermediate step between a machine and a human. Why do we need to tell the machine to show us what we need in a language we are supposed to understand? It would be better if the input was human language and output binarycode that the machine truly understand not just mimics predetermined code snippets.
So Software 3.0 is basically using the output of Software 2.0 and creating an infrastructure for it with Software 1.0. Wow, can't wait to see what Software 4.0 will be like.
U r not wrong, It is going to be bigger than all of those. And thats exactly what Jacque Fresco and The Venus Project has been saying since the 70s. And urged society to prepare for it in advance and shift the socio-economic model to accommodate that. Its time you listened. We need resource based economy. or at the very least a profit-free economy like Copiosis's or Common-Planet's Creditism.
I don't see how it's going to end up being as big as computers. I think it's here to stay, and is already useful, but i dont think it's going to be ubiquitous. A model that is useful just takes far too much compute. If someone invents a magical way to run a multi-billion parameter model on AA batteries, then yes, otherwise there will always be better ways to do lots of things
it's very important to understand that actual real people think simply adapting a foundational model to some random use case is a big thing. this is the Microsoft/Meta approach, use our model and ALSO all our tooling and social context for your business. i'm not sure how a little AI company without a massive social network and an army of implementation people can compete in software 3.0. this is the "sell picks and shovels to the gold miners" approach, which certainly worked for Levis
I am 95% sure that we will achieve AGI with multi-agent systems that work on multimodal language models and use tools, reflection, and sensors. There are still a few important atomic problems that GPT-4 cannot solve. To be 100% certain, I want to see that GPT-5 is indeed significantly smarter than GPT-4. If that is the case, I am very confident that we will create a completely new world with this technology.
Wes, your explanation of software 2.0 would have been more clear to me if you had mentioned objective functions as the way that we describe to the system what we want it to do. It's based on the result, and you tweak the weights until the result is what you want.
Biggest pain for the moment is training data formatting, you can’t just feed it a bunch of pdf and raw text, data has to be picked and formatted for training.
I have an idea that I think has potential but it requires dozens of AI agents, probably, to create a mockup to showcase it. It has to do with augmenting an existing thing. I have little energy due to a disorder, so I'm looking for others who may be interested to further develop the concept. We could talk about it sometime.
@@Laszer271 I'm looking to see if there's a way to basically enhance humanity's problem solving ability, to see if there is a way to help us solve the most urgent and important complex problems that we can't seem to solve any time soon. I think we're underutilizing experts and non-experts on current social media platforms. There are so many minds connected, but in impulsive and ineffective ways. I'm thinking of changing that and dramatically improving quality of communication and discussion. I've been looking a little into what we know about problem solving, and I think there may be ways to better utilize everyone's thoughts and traits. For instance, the way a problem is represented can drastically change how easy it is for someone to work on it. I also think that the most complex problems are so complex, that we don't have a good overview. I wonder if A.I. can have that overview where we cannot. There's a bit more to it, but that's the jist of it.
@@lystic9392 What? Dozens of agents? Nah, been there, done that, it's getting done in the industry. All depends on how you want to use them. But you have to take into account that agents generate cost, the more text they need to process the more cost there is. It's also important to consider what your target audience is. For average Joe spending $100 per hour for AI agents is unthinkable. For corpos, the same amount might be even too cheap as it is too good of a deal for them if it really provides a lot of value. The agent-architecture should be cheap (and simple) if you target average Joe but if you target corpos then it can be costly and advanced. All depends on the context of your project.
@@Laszer271 Sorry, RUclips made the original reply in which I describe my idea invisible. I don't know why. You have to click on 'newest comments' to see it. I'll try to post it again, see if RUclips allows for it.
Correct me if I'm wrong, but a SW3.0 implementation could also just be AutoGen and API calls to the latest greatest models too, if you can afford the token cost. No need to custom taior or teach a new model, just do that with a teachable agent.
If you can afford the token cost. At the moment we are getting heavily subsidised tokens. Subsidised by massive investments. This has to stop eventually. Small application specific models is a step towards this. That’s why every vendor is developing small models. To make their businesses sustainable. Once they can make money we will start to look at bigger models again to improve performance.
@@byrnemeister2008 We've been toying with the idea of really condensing the API calls and token count to the pay for play large models, and sort of "expert system'ing" a network of agents using a smaller open model. In my background of microcontroller firmware I find a good parallel - we regularly have to sift a signal out of analog noise. All of the basic algorithm tools and expert system syntax can really distill the prompt to a set of critical questions that can be asked to the pay for play models. That's my take so far. Working on something big, but the elephant in the room is "do I put the effort in, or wait for the big boys to just do it in a few years and prepare to work downstream from them?"
If AI Agents become the backbone of business/activity and the Internet goes down, ala CrowdStrike outage, we are actively building more and more "systems" on top of the Internet and just setting ourselves up for failure. I would imagine UNTIL we can make the Internet bullet proof, building more and more systems is a fool's errand. Civilization is based on Electricity and the Internet and if either of these goes down, MANY have no idea how to grow food, handle day to day life without these systems. VERY WORRIED this "Super Cool Futuristic AI Agents" is the normal evolution of apps/services but sure seems like a high deck of cards being added to the top of our very susceptible system and AI Agent's will one day actively take our systems DOWN!
Can we please get an AI that studies the constitution another AI studies the declaration of dependency what we’re doing wrong and then we need another bunch of ones interpret all of what’s going on in our country right now so we can start cracking down on people breaking the law
Billions of bots online, more bots than people, and remember most people don't produce anything online, they just follow, like and consume. The online space is going to be dominated by bot creators. Humans will be pushed even further to just absorbing 'persuasive' content. Where is this going?
It’s sort of like how the derivatives market is several times larger than all the actual financial assets including all stocks and holdings. It doesn’t really matter if the derivatives, these ethereal, abstract financial tools, become decoupled from the real assets they are supposed to be correlated to…. as long as they can be used to turn a profit. Similarly, it doesn’t matter if these bots mimicking humans have multiplied and run amok online…. as long as they can be used to turn a profit. So where is this all headed? These bots are going to have two purposes in regards to how they interact with humans: 1. They will try to get as much information as possible so as to help corporations extract as much money as possible from you with ads, products, etc. 2. They will actively mold human behavior so they can extract as much money as possible from you. They will try to keep you hooked and engaged so they can mold you better. This already the case with most of the algorithms that social media employs. They harvest your data, provide you targeted ads, and the advertising industry is actively trying to hijack your emotions. In the same way that the algorithms amplified the methodology from the TV era, the bots will just amplify the methodology from the algorithms.
The other thing that I think we should expect is that the bots will be given money so as to make more money. This will allow market manipulation on steroids. Imagine the kind of profits a hedge fund can make if they get an army of bots to create online hype about a certain stock or asset like a crypto currency by flooding comments and articles all over the internet and then they are able to get a bunch of AI agents to actually make it look like tons of people purchasing these assets. The ability to employ pump and dump schemes would be far, far more powerful than ever before. People will be able to make literally over a billion dollars doing this kind of thing. With incentives like that, it makes this outcome almost completely inevitable.
If context windows are going up and inference prices are going down, Why do we need to fine-tunning small models in the future? isn't better to do incontext learnig? we'll simply use a System Prompt to make an agent. Sorry, I dont see the bussiness.
Wes really has the ML crowd spirit. Quantity over Quality!! More and longer videos with less and less information or eddeting. Very lazy content TBH. The only thing im giving you is that you're always up to date and therefore pop up in my YT feed first.
I disagree that technical expertise will not be required for software 3.0 As someone who works in both Data Engineering and Data Science, we will always need engineers to help build out the infrastructure that Scientists, Analysts and Users use to do their work. There will always, ALWAYS be a need for someone to “look under the hood” and work in code, work with the physical hardware and produce the physical hardware. Look at the Windows bug that shut down global transportation a week ago. This was due to some bad kernel level code, done by an engineer who didn’t fully understand the deep software issue. If anything, the demand for technical expertise will increase and those technical will benefit the most be having their work being super charged by millions of AI agents. Also the biggest money maker here is starting an agency that helps businesses integrate software 3.0 into their current business. That is the shovel in this gold rush.
Folks like that might exist at the forefront of technological gaps for the next several years. Once these models have context windows that amount to self learning capabilities those gaps will shrink to the point of near non-existence. So yeah, those folks might always be needed but they go from millions of people in the USA to a thousand or two across the whole planet. About the same number of people who write entries for RSA competitions.
My dude, Have you completely missed the robots that are being built with AI running them? Even today companies are already in the beginning phases of replacing their workers with robots. Amazon is now officially buying more robots to work in its warehouses than it's hiring new employees. Tesla is deploying their Optimus robots in their factories. No, companies will NOT always need someone to look under the hood. We're closing in on having AI do both software and hardware. ---- The Gospel of Lord Jesus John 3:16-18 (Jesus speaking) 16 For God so loved the world, that he gave his only begotten Son, that whosoever believeth in him should not perish, but have everlasting life. 17 For God sent not his Son into the world to condemn the world; but that the world through him might be saved. 18 He that believeth on him is not condemned: but he that believeth not is condemned already, because he hath not believed in the name of the only begotten Son of God. Romans 10:9 That if thou shalt confess with thy mouth the Lord Jesus, and shalt believe in thine heart that God hath raised him from the dead, thou shalt be saved. Repent of your sins or suffer the consequences. Lord Jesus died in our places personally to take the death punishment that sin deserves and then resurrected by the power of God. Believe this and sincerely repent of your sins each time you sin and you will have eternal life and nothing to fear. Fail to repent and you will end up in the Lake of Fire.
A lot of these videos are being made by non technical people and is extremely misleading. In the comment section here, some guy is quoting a bible verse. Why? This whole “AI Boom” is attracting the worst kinds of individuals. People are going to treat this like a religion and assume AI will take care of everything. That is just being willingly ignorant.
@@AaronAsherRandall we have systems like Ansible where one admin can maintain nearly an unlimited numbers of machines. The biggest limiting factor with that is just the admin knowing how to get started. To move away from server side, previously folks asked their friend / family member whatever question they had on how to fix something. There is no reason an AI can't handle those questions faster and better. Those people make up the vast majority of those millions of technical support people that exist today. The skill level to meaningfully provide support in twenty years is going to be absurd due to LLMs. Next 2 years you're absolutely right, we won't be there yet, but I wasn't talking about the near term tomorrow.
The world of AI framworks is facinating to me, since u can mix and match agents with different capabilities along the cost, speed and accuracy curve. So u may have free AI such as LAMA 3.1 doing the grunt general purpose work that is needed most of the time. Whilst u use specialist AI agents that are expensive but they are experts, highly accurate and fast as fuck. The teacher LAMA's are also needed for when the framework draws a conclusion from partial products produced or obtained by the specific agents. If the framework has a negative or positive feeback or feedforward configuation then the desired result can be compared with the partial product. These itative comparison can drive the teacher AI ti say ok, these are the errors or difference between desired partial output what do need to do to remove or minimise the difference. The teacher will then say, OK what kind of AI do need to train or how can I train an existing AI that I tought before, better. This is very interesting since there is reaning and understanding that AI will get things wrong. Even the most expert of expert AI will get some things wrong relstive to what we want or logically wrong. To be honest having a Formal Maths AI or classical Formal reasoning machine as part of thid framework would be great. This woulf ensure thay any step could be checked for jt logical behsvious was ensured.
Who are the customers? Your customers used to be next door company employees. And next door company customers used to be your employees. If there are no employees anymore then who are the customers? UBI? With whose money? Competition will drive the value of intellectual services down to the cost of electricity. A very few large winners will offer all the intellectual services for dirt cheap because that's the amount of money that the remaining "customers" can afford. Software 3.0 will be a huge disruption and destruction of value.
If you or Mark Zuckerberg is right and there will be a day when we can spin up AI in a few clicks. If it’s so easy that a child can spin up AI for some Pokémon. There is literally no monetary value for any entrepreneur out there. As fascinating as I think AI is, I can’t help but think that all we’re doing is creating technologies That’ll create a future where big businesses don’t really need people anymore. You’re either working on AI for minimum wage, or you’re laid off. I’m not a medium, and I hope I’m wrong. But that’s what it feels like.
You're not alone in thinking that my guy. I feel it's going to do nothing but rocket us towards a dystopian future where the sweat of a man's brow is rendered obsolete. But with no one to be able to consume said products, then there will be no big business either, unless Amazon or meta decides to just make a governmental coup a reality and place themselves at the helm of our big brother government, but I digress.
"There is literally no monetary value for any entrepreneur out there." - strongly disagree, there is more value than ever before actually. Entrepreneurship isn't just about creating AI, it's about using it to optimize processes and open new, previously unachievable avenues. Using it very cleverly with human-in-the loop or building an IT infrastructure that AI can leverage will be some of the things that will give you an edge.
There's business outside of digital work. Fun-fact: Human labor can be much more energy efficient than using machines of any kind. Even for crushing stones into pebbles. 😅
Phenomenal heads-up! Pairing these bottled knowledge sub-model agents with iot and block chains where sensor data is translated into business data, (this data is based on actual moment-in-time business sensor data which the ai makes usable). Accessing and making actionable business decisions using sensors psired with ai, which is then recorded in real-time on industry or company specific distributed ledgers (smart coins) seems to have billions of $ in opportunities. The future is now. And now ;-)
Dear Mr Wes. I like your content. Srsly. But no one who watches your videos needs explaining the basics of computing or maschine learning. We didn't get here from "10 best movies with Rayan Gosling" by Watch Mojo or 5 ways to clean a car. We got here because we love to hear about AI and stuff. Lil respect for the audience my man!
God, for this price, I expect to have ai that can make full aaa games, or for 600 billion, you're getting ripped off so final fantasy 16 tales of arise god of war red dead redemption and gta 6 for that price
So many assumptions…yet ignoring the obvious. The massive GPU power resources on such a ridiculous scale…will come from where? Last time I checked electrical power is insanely expensive. On top of that current GPU architecture is incredibly sloppy both in terms of power consumption and heat, regardless of the $value of Nvidia. The premise of LLM’s storing extremely large datasets…is a stone age problem definitely NOT ready global release for the masses. Ease up everyone, and think realistically.
Oh, Mark's trying hard to hold onto Capitalism. Telling all those small businesses how they can thrive. Or is it just a COVER (me thinks so). If they can't see socialism at this point, they wouldn't have made it past 3rd grade...
I think a lot of your thoughts make sense, but in general we should have low confidence that we can predict the future of technology and how it's used. Things that make a lot of sense right now might begin not to in short timelines.
90% of AI is just hype. Altought there are few good use cases, but it will not change Humanity. We are obsessed with benchmarking new models, but getting nothing out of it. Yes AI can write code, make robot walk few steps etc. but it's not revolutionary.
I'm glad you told me. So it looks like Sundar Pichai, Satya Nadella, Mark Zuckerberg, Elon Musk, Jensen Huang , Demis Hassabis, Ilya Sutskever, Yann LeCun, Andrew Ng, Dr.Fei Fei Li, Geoffrey Hinton, Yoshua Bengio, Mustafa Suleyman etc. have been gaslighting us all along.
Very insightful video, I remember Andrej covering something similar in his video "Intro to LLM's". He referred to the new age of systems as LLM OS, but after this video, it seems he was referring to Software 3.0. Especially after seeing new frameworks like Agent Zero that allow for an agent to basically build out anything it needs, solve errors it encounters, It can delegate tasks to other agent instances, it has access to tools etc.
The entire concept of running a business with just yourself and AI agents are very interesting and seem lucrative but the thing I can’t wrap my mind around is - if everyone does this, its absolutely no way our world wide economy can withstand it. This will most definitely, imho, cause the emergence of a new type of economic model that will be a successor to Capitalism.
Maybe a new economic model that is compartmentalized in that each market is gauged or valued based upon how their respective businesses help push, advance and evolve the market it is within, for example;
A customer service compartment may raise or lower the value of a AI-Human hybrid company based upon how their practices help better or lower the quality of Customer Service.
Things are about to change and I think for the better
Exactly. There is a finite amount of wealth to capture in the world. What happens when everybody has the intelligence and capability to start their own business? Nobody wins in the end, except the legacy big players who already have accumulated wealth.
so capitalism won't find a workaround to embed itself into new technology? first time?
Capitismentism
That is the point. The current system needs to break. It never worked for the majority. Nothing new can come until it does break.
Just let this blow your mind first.
What do you wish for first for with a genie that has no limitations on wishes.
You wish for infinite wishes.
What do you do when you get your own ai driven robot.
You obviously have it make more robots.
Then you sell the robots that are made.
Or lets try something a bit different.
Lets say you tell your ai robot you need it to make a farm that would fit into a certain space (indoors or outdoors) give it specifics on caloric needs for your family monthly and have it build it and set it up as a closed system so you have most of your caloric intake accounted for year round. Now what happens when u excitedly tell your neighbors and they want it too. U lend em ur robot, and you now take out a neighborhoods worth of capital out of grocery stores, then what happens to the truckers, what happens to the farmers, what happens to all the downstream ripples caused by one action?
New actions breed new economies, but we will be economies of scale, trading in real time commodity trading on local block chains.
Imagine you specialize in tool making because you like the idea of working a forge, and instead you sell your tools on the block chain, then you take your "earnings" and then get food you need from your neighbor down the street who likes to farm. You get your entertainment locally, from the creative people all around you.
People will always want to work when it is doing something they love. And this system will free you up to follow what it is you really want to do. A golden age is coming. Its not that far off, maybe 2~3 years 🤔
It's crazy that this video has 14 minutes of yapping before saying what 'software 3.0' is.
Thanks for the heads up, very useful
@@Mrv958Yup. If anyone wants to save their time, software 1.0 is a person writing a computer program for a specific task. Software 2.0 is a person transcending the need to program by training a specialized AI to do that same task more efficiently. Software 3.0 is a big, general, and open source AI (ie llama 3.1 large) that trains specialized AI for specific tasks. In other words, a distinction for distinction's sake.
Context is important. Just google it if you want an immediate surface-level answer
Thank you
Thank you!
check out the game Screeps, it's a game designed to teach Java Script, but because the community has documented everything for years, Chat GPT can just code it amazingly well right out of the box
Finally, a way to play screeps HAHA
This will be much bigger than any other business wave. The crazy thing is the foundations are here. People just don’t know how to use them. As a consultant and university educator, I tell people- the world has already moved your cheese, you just don’t know it! Scary for many. Exciting for some. Well done!
Not sure Software 3.0 will be big models training small models.
I mean, yes. We will see a lot of that. But i don't see it as a step change up from 2.0, as 2.0 was a step change from 1.0.
Karpathy himself warned about the diminishment of model effectiveness when overwriting, or abstracting from, model weights. If you want a better model, you retrain it with new data *and* old data. Get weights right for the whole enchilada.
If you abstract a smaller model from a larger one, will there be a bias towards the mean? A lesser field of competence?
I dunno.
I've been thinking instead of 3.0 as self-improving, self-actualizing, self-optimizing models. Q-Star, perhaps. A little furthur down that path is AGI.
I don't think this two-tiered model development idea is on that path. But maybe I'm wrong?
Whether it ought to be called Software 3.0 or something different, I am thinking along similar lines in that I think a key component of AGI will be the ability to learn on the fly (self-improvement), among the others you mention.
"I've got to warn you, it's a learning robot. Every moment you spend fighting it only increases its knowledge of how to beat you."
Every piece of useful software you've ever interacted with is a result of careful planning and hard work of very smart problem solvers. What we currently call AI is just a fancy auto-complete engine. Just remember that.
This is absolutely insane. End of the year this bubble will burst.
Great video.❤ I work so hard to engage myself in someway of earning more income and stop depending on anyone. my family are happy once again and I can now afford anything for my family even with my Retirement. With a step of $50k and $17k weekly returns has been life changing, after so much struggles.
Hello how do you make such weekly??
I'm a born Christian and sometimes I feel so down of myself because of low finance but I still believe in God.
@@babytruca02Maria Angelina Alexander I really appreciate her efforts and transparency.
My outlook on money changed when I realized someone making $200K can retire broke & someone making $80K can retire a millionaire. With the current market movement.
You're correct!! I make a lot of money without relying on the government.
Investing in stocks and digital currencies is beneficial at this moment.
wow this awesome 👏 I'm 37 and have been looking for ways to be successful, my work hasn’t been helpful.
Zuck looks like Merry Poppins after hard night drinking.
Yea, I agree with the other posts, I love Wes’s videos but sometimes I get hung up on how long they are. Maybe he needs to hire an editor to help? I’d be curious what would happen if he took his own transcript as a rough draft and put it into ai to make it more succinct, then edit as necessary following that. I bet the results would be amazing.
Definitely not crazy; it all depends on the cost of running those product models. We simply can't have that trivia if the mean medium cost are to thousands of dollars a day!
Thanks for the video! ❤
Zuck has dethroned them all... EPIC! 🙏 This could be one of the biggest PR turnarounds in modern business. Great job Wes, framing stories that matter!
2:35 Hebbian learning (neurons that fire together, wire together) isn't much used for current _artificial_ intelligence. Instead, so-called backpropagation is the predominant algorithm for training AI. For an overview, see Figure 1 in "Backpropagation and the brain" by Geoffrey Hinton and coauthors.
Thank you Lord Jesus for the gift of life and blessings to me and my family $14,120.47 weekly profit Our lord Jesus have lifted up my Life!!!🙏❤️❤️
I'm 37 and have been looking for ways to be successful, please how??
Sure, the investment-advisor that guides me is..
Mrs Kathy lien
Her services is the best, I got a brand new Lambo last week and paid off my mortgage loan thanks to her wonderful services!
😱Sounds familiar, I have heard her name on several occasions.. and both her success stories in the wall Street journal!
so if you didn't have enough time to deal with all the bots on Facebook, Mark suggests more bots!
My agent would simply scrape what I wanted off of Facebook, like just posts from people I follow. No ads, no nothing and very little info given to the zuck. Same for Google searches.
One of the hallmarks of mania is a rapid rise in complexity.
AI has become a giant bubble waiting to pop. Most of this technology is overhyped in its utility, and the profitably of that which is useful is dubious, because the ways it’s leveraged are democratized and can be done for free, and the inferencing is highly competitive.
Which is why actually learning how to use it, and create things for yourself is the real value and profit opportunity.
It's crazy good if you know how to use it, so the tech itself is not overhyped, I would dare to say it's even underhyped. What is overhyped though are some AI-based products like Devin. AI itself won''t pop, waiting for it to pop is like waiting for computer or phone or internet "bubble" to pop - not gonna happen.
@@Laszer271 The way some people are talking about AI now is the way that I remember some people talking about the Internet in the 90s - it's overblown, it will pop, any day now... Well, the tech bubble popped in 2000, but that was a miniscule bubble compared to where tech is now. And the Internet just kept exploding.
Most people and organizations are just learning how to use/integrate AI. It's underhyped. The results in medical companies that have done so (with AlphaFold etc.), from two whole years of integration, have produced astounding results.
So yeah, even the models we have now like GPT-4 or Claude Sonnet are enough to drastically change society over the next decade, and the models just keep getting better.
@@squamish4244 Completly agree. AI right now is sports car in the middle of the deep forest. People found it and are complaining that it can't do much and walking is much more efficient. They can't imagine that the infrastructure for the car to really take off is only just being built.
GPT-5 is not needed, we don't really need faster car, all we need are roads.
@@Laszer271 Speaking of cars, don't even get me started on how some people think that AI and self-driving cars are a good analogy, and "Look at self-driving cars, we don't have them yet, therefore AI is overhyped." There are so many problems with that analogy, good lord!
GPT-5 is coming anyway, and LLMs went from "amazing" to "meh" in AI popular discourse in less than two years, but they're already moving past LLMs (LLMs only), so we have A LOT of building out of infrastructure to do in the next decade.
I did not expect the Factorio at the end of the video Hahaha
Nice one, Wes Roth!
Love your philosophical takes! Do you think it will be more procedural generated? Software does not have to go away.. We need a workbench (comadore amiga reverence) for certain tasks right.. A photoshop 3.0 (or 2.0 for we are still stuck in a program from 1990) we will give a command and the really small procedural agents do this tasks.. We will guide them.. Little bit to the left or right give guidance, but then it will be more like clay.. But the workbench still needed right? Not all is abstract thinking in words...
Love your work... respect!
Thank you for such increible information and explanation and I share your thoughts about the future, my opinion would be even bigger that we expect
Yep, I'll keep repeating it over and over. We can have a manageable, Zero-Loss AI transition through both sensible regulation and by the fact that AI is going to enable MORE jobs than it would have replaced. We have years of work converting not just hearts n minds to AI use, but businesses themselves. Use cases to be discovered, data hoards unleashed from private companies. I know of two geologists who are building AI based on centuries of company mining data, and they're gonna be able to send teams out and pinpoint sample pattern areas. Success means more jobs especially if the company hits paydirt at multiple sites.
My kids are already doing that ;)
Thank you.
Awesome channel dude! Thank you 🙏🏻 #respek
5:56 You wondered if anyone would catch the Speed reference. I did, hotshot.
To consider: programming language is a intermediate step between a machine and a human. Why do we need to tell the machine to show us what we need in a language we are supposed to understand? It would be better if the input was human language and output binarycode that the machine truly understand not just mimics predetermined code snippets.
So Software 3.0 is basically using the output of Software 2.0 and creating an infrastructure for it with Software 1.0. Wow, can't wait to see what Software 4.0 will be like.
That's literally the point of iteration. 2.0, 3.0, 3.5... Or else it would be "New Thing" 1.0
But what is the conversion software version 7.0?
I will unsuncribe if you pull stuff like this again. Supposedly building tension to say something underwhelming. Just say it.
Why not just run the transcription through an LLM if you are on a time crunch?
@@Loflou Id rather unsubscribe than make a workaround for an issue he created. There are alternatives.
Hahaha, you funny. You are also totally irrelevant.. but funny.
Ohhh we will miss u....Not lol
U r not wrong, It is going to be bigger than all of those.
And thats exactly what Jacque Fresco and The Venus Project has been saying since the 70s.
And urged society to prepare for it in advance and shift the socio-economic model to accommodate that.
Its time you listened.
We need resource based economy. or at the very least a profit-free economy like Copiosis's or Common-Planet's Creditism.
Black Box will happen sooner than later, but we have to scale the Trust hurdle first. 🙏
I don't see how it's going to end up being as big as computers. I think it's here to stay, and is already useful, but i dont think it's going to be ubiquitous. A model that is useful just takes far too much compute. If someone invents a magical way to run a multi-billion parameter model on AA batteries, then yes, otherwise there will always be better ways to do lots of things
is software 3.0 basically chatGPT's custom models?
19:25 - what hole do I live in!? Iv been waiting for someone to agentise Factorio this entire time! Thanks for the heads up! ❤
it's very important to understand that actual real people think simply adapting a foundational model to some random use case is a big thing. this is the Microsoft/Meta approach, use our model and ALSO all our tooling and social context for your business. i'm not sure how a little AI company without a massive social network and an army of implementation people can compete in software 3.0. this is the "sell picks and shovels to the gold miners" approach, which certainly worked for Levis
tidak akan pergi Not Never
Thought provoking. Thank you.
I am 95% sure that we will achieve AGI with multi-agent systems that work on multimodal language models and use tools, reflection, and sensors. There are still a few important atomic problems that GPT-4 cannot solve. To be 100% certain, I want to see that GPT-5 is indeed significantly smarter than GPT-4. If that is the case, I am very confident that we will create a completely new world with this technology.
Is this because of sun glasses, or because of the meta quest? I reference to the bright eye sockets.
The moment people stop programming themselves will be the beginning of the end for us.
Wes, your explanation of software 2.0 would have been more clear to me if you had mentioned objective functions as the way that we describe to the system what we want it to do. It's based on the result, and you tweak the weights until the result is what you want.
Mark is a most NPC looking person in the world.
Biggest pain for the moment is training data formatting, you can’t just feed it a bunch of pdf and raw text, data has to be picked and formatted for training.
Keep in mind that even something like chatGPT's frontend was written by human software developers. This is nothing more than a VC pitch.
I have an idea that I think has potential but it requires dozens of AI agents, probably, to create a mockup to showcase it. It has to do with augmenting an existing thing.
I have little energy due to a disorder, so I'm looking for others who may be interested to further develop the concept. We could talk about it sometime.
Dozens of agents is easy but also costly. What do you need it for?
@@Laszer271 I'm looking to see if there's a way to basically enhance humanity's problem solving ability, to see if there is a way to help us solve the most urgent and important complex problems that we can't seem to solve any time soon.
I think we're underutilizing experts and non-experts on current social media platforms. There are so many minds connected, but in impulsive and ineffective ways. I'm thinking of changing that and dramatically improving quality of communication and discussion.
I've been looking a little into what we know about problem solving, and I think there may be ways to better utilize everyone's thoughts and traits. For instance, the way a problem is represented can drastically change how easy it is for someone to work on it. I also think that the most complex problems are so complex, that we don't have a good overview. I wonder if A.I. can have that overview where we cannot. There's a bit more to it, but that's the jist of it.
@@Laszer271 Or does that sound a little bit too wild?
@@lystic9392 What? Dozens of agents? Nah, been there, done that, it's getting done in the industry. All depends on how you want to use them.
But you have to take into account that agents generate cost, the more text they need to process the more cost there is. It's also important to consider what your target audience is. For average Joe spending $100 per hour for AI agents is unthinkable. For corpos, the same amount might be even too cheap as it is too good of a deal for them if it really provides a lot of value.
The agent-architecture should be cheap (and simple) if you target average Joe but if you target corpos then it can be costly and advanced. All depends on the context of your project.
@@Laszer271 Sorry, RUclips made the original reply in which I describe my idea invisible. I don't know why. You have to click on 'newest comments' to see it.
I'll try to post it again, see if RUclips allows for it.
Correct me if I'm wrong, but a SW3.0 implementation could also just be AutoGen and API calls to the latest greatest models too, if you can afford the token cost. No need to custom taior or teach a new model, just do that with a teachable agent.
If you can afford the token cost. At the moment we are getting heavily subsidised tokens. Subsidised by massive investments. This has to stop eventually. Small application specific models is a step towards this. That’s why every vendor is developing small models. To make their businesses sustainable. Once they can make money we will start to look at bigger models again to improve performance.
@@byrnemeister2008 We've been toying with the idea of really condensing the API calls and token count to the pay for play large models, and sort of "expert system'ing" a network of agents using a smaller open model. In my background of microcontroller firmware I find a good parallel - we regularly have to sift a signal out of analog noise. All of the basic algorithm tools and expert system syntax can really distill the prompt to a set of critical questions that can be asked to the pay for play models. That's my take so far. Working on something big, but the elephant in the room is "do I put the effort in, or wait for the big boys to just do it in a few years and prepare to work downstream from them?"
Dystopian.
If AI Agents become the backbone of business/activity and the Internet goes down, ala CrowdStrike outage, we are actively building more and more "systems" on top of the Internet and just setting ourselves up for failure. I would imagine UNTIL we can make the Internet bullet proof, building more and more systems is a fool's errand. Civilization is based on Electricity and the Internet and if either of these goes down, MANY have no idea how to grow food, handle day to day life without these systems. VERY WORRIED this "Super Cool Futuristic AI Agents" is the normal evolution of apps/services but sure seems like a high deck of cards being added to the top of our very susceptible system and AI Agent's will one day actively take our systems DOWN!
I believe the first AGI will have been written by other AI.
Can we please get an AI that studies the constitution another AI studies the declaration of dependency what we’re doing wrong and then we need another bunch of ones interpret all of what’s going on in our country right now so we can start cracking down on people breaking the law
Probably going after the 1% that are actively breaking a lot making life miserable for the 99%
Damn, i thought the buzzwords would kill me in devops, nothing like ai.
Billions of bots online, more bots than people, and remember most people don't produce anything online, they just follow, like and consume. The online space is going to be dominated by bot creators. Humans will be pushed even further to just absorbing 'persuasive' content. Where is this going?
It’s sort of like how the derivatives market is several times larger than all the actual financial assets including all stocks and holdings. It doesn’t really matter if the derivatives, these ethereal, abstract financial tools, become decoupled from the real assets they are supposed to be correlated to…. as long as they can be used to turn a profit. Similarly, it doesn’t matter if these bots mimicking humans have multiplied and run amok online…. as long as they can be used to turn a profit.
So where is this all headed?
These bots are going to have two purposes in regards to how they interact with humans:
1. They will try to get as much information as possible so as to help corporations extract as much money as possible from you with ads, products, etc.
2. They will actively mold human behavior so they can extract as much money as possible from you. They will try to keep you hooked and engaged so they can mold you better.
This already the case with most of the algorithms that social media employs. They harvest your data, provide you targeted ads, and the advertising industry is actively trying to hijack your emotions. In the same way that the algorithms amplified the methodology from the TV era, the bots will just amplify the methodology from the algorithms.
The other thing that I think we should expect is that the bots will be given money so as to make more money. This will allow market manipulation on steroids. Imagine the kind of profits a hedge fund can make if they get an army of bots to create online hype about a certain stock or asset like a crypto currency by flooding comments and articles all over the internet and then they are able to get a bunch of AI agents to actually make it look like tons of people purchasing these assets. The ability to employ pump and dump schemes would be far, far more powerful than ever before. People will be able to make literally over a billion dollars doing this kind of thing. With incentives like that, it makes this outcome almost completely inevitable.
If context windows are going up and inference prices are going down, Why do we need to fine-tunning small models in the future? isn't better to do incontext learnig? we'll simply use a System Prompt to make an agent.
Sorry, I dont see the bussiness.
Wes really has the ML crowd spirit. Quantity over Quality!!
More and longer videos with less and less information or eddeting. Very lazy content TBH. The only thing im giving you is that you're always up to date and therefore pop up in my YT feed first.
I disagree that technical expertise will not be required for software 3.0
As someone who works in both Data Engineering and Data Science, we will always need engineers to help build out the infrastructure that Scientists, Analysts and Users use to do their work. There will always, ALWAYS be a need for someone to “look under the hood” and work in code, work with the physical hardware and produce the physical hardware.
Look at the Windows bug that shut down global transportation a week ago. This was due to some bad kernel level code, done by an engineer who didn’t fully understand the deep software issue.
If anything, the demand for technical expertise will increase and those technical will benefit the most be having their work being super charged by millions of AI agents.
Also the biggest money maker here is starting an agency that helps businesses integrate software 3.0 into their current business. That is the shovel in this gold rush.
Folks like that might exist at the forefront of technological gaps for the next several years. Once these models have context windows that amount to self learning capabilities those gaps will shrink to the point of near non-existence. So yeah, those folks might always be needed but they go from millions of people in the USA to a thousand or two across the whole planet. About the same number of people who write entries for RSA competitions.
My dude, Have you completely missed the robots that are being built with AI running them? Even today companies are already in the beginning phases of replacing their workers with robots. Amazon is now officially buying more robots to work in its warehouses than it's hiring new employees. Tesla is deploying their Optimus robots in their factories.
No, companies will NOT always need someone to look under the hood. We're closing in on having AI do both software and hardware.
----
The Gospel of Lord Jesus
John 3:16-18 (Jesus speaking)
16 For God so loved the world, that he gave his only begotten Son, that whosoever believeth in him should not perish, but have everlasting life.
17 For God sent not his Son into the world to condemn the world; but that the world through him might be saved.
18 He that believeth on him is not condemned: but he that believeth not is condemned already, because he hath not believed in the name of the only begotten Son of God.
Romans 10:9
That if thou shalt confess with thy mouth the Lord Jesus, and shalt believe in thine heart that God hath raised him from the dead, thou shalt be saved.
Repent of your sins or suffer the consequences. Lord Jesus died in our places personally to take the death punishment that sin deserves and then resurrected by the power of God. Believe this and sincerely repent of your sins each time you sin and you will have eternal life and nothing to fear. Fail to repent and you will end up in the Lake of Fire.
A lot of these videos are being made by non technical people and is extremely misleading. In the comment section here, some guy is quoting a bible verse. Why? This whole “AI Boom” is attracting the worst kinds of individuals. People are going to treat this like a religion and assume AI will take care of everything. That is just being willingly ignorant.
@@AaronAsherRandall we have systems like Ansible where one admin can maintain nearly an unlimited numbers of machines. The biggest limiting factor with that is just the admin knowing how to get started. To move away from server side, previously folks asked their friend / family member whatever question they had on how to fix something. There is no reason an AI can't handle those questions faster and better. Those people make up the vast majority of those millions of technical support people that exist today. The skill level to meaningfully provide support in twenty years is going to be absurd due to LLMs. Next 2 years you're absolutely right, we won't be there yet, but I wasn't talking about the near term tomorrow.
ACCELERATE
The world of AI framworks is facinating to me, since u can mix and match agents with different capabilities along the cost, speed and accuracy curve.
So u may have free AI such as LAMA 3.1 doing the grunt general purpose work that is needed most of the time. Whilst u use specialist AI agents that are expensive but they are experts, highly accurate and fast as fuck. The teacher LAMA's are also needed for when the framework draws a conclusion from partial products produced or obtained by the specific agents.
If the framework has a negative or positive feeback or feedforward configuation then the desired result can be compared with the partial product. These itative comparison can drive the teacher AI ti say ok, these are the errors or difference between desired partial output what do need to do to remove or minimise the difference.
The teacher will then say, OK what kind of AI do need to train or how can I train an existing AI that I tought before, better.
This is very interesting since there is reaning and understanding that AI will get things wrong. Even the most expert of expert AI will get some things wrong relstive to what we want or logically wrong.
To be honest having a Formal Maths AI or classical Formal reasoning machine as part of thid framework would be great. This woulf ensure thay any step could be checked for jt logical behsvious was ensured.
"I said this before in the future." - Wes Roth (quoted out of context)
Software 1.5: DevOps.
integrating it into other software been challenge for me, doing API, etc... hard... i was like interpreter idea...
What?
@@honkytonk4465 sorry I was thinking my experience with agents silly me commenting my self
Yeah but that's 14 minutes and he didn't cry
that was a very long way to explain that there will be "templates"
Can someone timestamp 3.0
Who are the customers?
Your customers used to be next door company employees. And next door company customers used to be your employees.
If there are no employees anymore then who are the customers?
UBI? With whose money?
Competition will drive the value of intellectual services down to the cost of electricity. A very few large winners will offer all the intellectual services for dirt cheap because that's the amount of money that the remaining "customers" can afford.
Software 3.0 will be a huge disruption and destruction of value.
Very interesting 🤔🤔😎
If you or Mark Zuckerberg is right and there will be a day when we can spin up AI in a few clicks. If it’s so easy that a child can spin up AI for some Pokémon. There is literally no monetary value for any entrepreneur out there. As fascinating as I think AI is, I can’t help but think that all we’re doing is creating technologies That’ll create a future where big businesses don’t really need people anymore. You’re either working on AI for minimum wage, or you’re laid off. I’m not a medium, and I hope I’m wrong. But that’s what it feels like.
You're not alone in thinking that my guy. I feel it's going to do nothing but rocket us towards a dystopian future where the sweat of a man's brow is rendered obsolete. But with no one to be able to consume said products, then there will be no big business either, unless Amazon or meta decides to just make a governmental coup a reality and place themselves at the helm of our big brother government, but I digress.
"There is literally no monetary value for any entrepreneur out there." - strongly disagree, there is more value than ever before actually. Entrepreneurship isn't just about creating AI, it's about using it to optimize processes and open new, previously unachievable avenues. Using it very cleverly with human-in-the loop or building an IT infrastructure that AI can leverage will be some of the things that will give you an edge.
There's business outside of digital work. Fun-fact: Human labor can be much more energy efficient than using machines of any kind. Even for crushing stones into pebbles. 😅
@@Laszer271 Ah, I didn’t think about it that way.
Phenomenal heads-up!
Pairing these bottled knowledge sub-model agents with iot and block chains where sensor data is translated into business data, (this data is based on actual moment-in-time business sensor data which the ai makes usable).
Accessing and making actionable business decisions using sensors psired with ai, which is then recorded in real-time on industry or company specific distributed ledgers (smart coins) seems to have billions of $ in opportunities.
The future is now. And now ;-)
Mark sounds like hes drowning in his saliva Truly the pavlovian response is on display here.
a whole lot of small caps and IPOs will come from AI agents building and running businesses.
It’s time to go home when you start to look like your metaverse avatar. 😂
v4.0 will be neural link ;)
now we need an open source NPU or GPU or LPU or what ever :D
i mean share this technology to every body, or at least sell them to the public.
So, 3.0 = A.I.-led fine tuning?
Ai write Spaggetty code , that gone generate a Lots of problems
Dear Mr Wes. I like your content. Srsly. But no one who watches your videos needs explaining the basics of computing or maschine learning. We didn't get here from "10 best movies with Rayan Gosling" by Watch Mojo or 5 ways to clean a car. We got here because we love to hear about AI and stuff. Lil respect for the audience my man!
God, for this price, I expect to have ai that can make full aaa games, or for 600 billion, you're getting ripped off so final fantasy 16 tales of arise god of war red dead redemption and gta 6 for that price
So many assumptions…yet ignoring the obvious. The massive GPU power resources on such a ridiculous scale…will come from where? Last time I checked electrical power is insanely expensive. On top of that current GPU architecture is incredibly sloppy both in terms of power consumption and heat, regardless of the $value of Nvidia. The premise of LLM’s storing extremely large datasets…is a stone age problem definitely NOT ready global release for the masses. Ease up everyone, and think realistically.
So we can Pavlov’s dog (training style) to hack Ai
So....web 3? software is only a section of the web.
These are not weights. What kind of people are you? These are synapses.
The most ShOcking part of the video is a raccoon impersonating Mark Zuckerberg...
Is software 3.0 akin to Web 3.0 a few years ago? 😂
OMG your diagrams Lolz
When aponia is upon you there is pleasure to mention. Uh uh uh! Look who wants to teach me quantum physics...the kind that you humans never listened.
Looks like a trap. Don’t fall for it. Cease Ai. Fight back.
Zuk has been out surfing his private beech for the last month. Not a good look
Oh, Mark's trying hard to hold onto Capitalism.
Telling all those small businesses how they can thrive.
Or is it just a COVER (me thinks so).
If they can't see socialism at this point, they wouldn't have made it past 3rd grade...
I think a lot of your thoughts make sense, but in general we should have low confidence that we can predict the future of technology and how it's used. Things that make a lot of sense right now might begin not to in short timelines.
❤❤🎉🎉❤❤🎉🎉❤❤
90% of AI is just hype. Altought there are few good use cases, but it will not change Humanity. We are obsessed with benchmarking new models, but getting nothing out of it. Yes AI can write code, make robot walk few steps etc. but it's not revolutionary.
yup.nothing new under the sun.
Yeah, overhyped like the printing press.
You're blending opinions of experts with those of bean counters. You're diluting genius juice with corn syrup. It's not healthy.
Additionally someone needs to tell the Zuck that its important to tan around your eyes as well. Otherwise ya look like a takooni
Congratulations, I unsubscribed.
Boring demonstration 👎
AI bubble will burst soon.
I'm glad you told me. So it looks like Sundar Pichai, Satya Nadella, Mark Zuckerberg, Elon Musk, Jensen Huang , Demis Hassabis, Ilya Sutskever, Yann LeCun, Andrew Ng, Dr.Fei Fei Li, Geoffrey Hinton, Yoshua Bengio, Mustafa Suleyman etc. have been gaslighting us all along.
Second viewer! :D