As a Brazilian The prices of Open IA is really prohibitive because the Dollar is too expensive. DeepSeek will really OPEN the IA for Hundred of countries. Its a trully Game Changer
The China point is so much more nuanced than what's presented here. Yes there is stealing, mimicry, copying, etc. But look at the surnames on many of the research papers pushing the frontier forward. The Chinese academic system is massive and is churning out unprecedented amounts of bona fide AI researchers and engineers. People who are incredibly smart and have the State apparatus posturing unparalleled resources to back them. They shouldn't be underestimated.
Correct! I saw a research paper the other day. I cant remember exactly what paper it was, but it had 14 researchers on board. 13 Chinese and one middle-eastern. M$ or something..It's the same all over, especially in AI.
100% since they dont have the same kind of regulations they can just build endless coal power plants to produce their power. And let's be real, Asian cultures prioritize education and achievement. If you're great at STEM in the states you're getting bullied. Congrats on getting new dogs! If my cat doesn't knock something off my desk or beg loudly for treats during work what is even life? Having office animals is the best.
It's reasoning process is the most sophisticated I've seen to date. And very human like in the way it questions itself. This company doesn't appear to be behind OpenAI at all. I'd say it's ahead, when taking into consideration the size of the models.
I love the transparency and thinking process of it. It’s fun to try to play chess with it even if it forgets what’s going on because you can see how it thinks. It’s hilarious when it makes illegal moves, but educational to see the thought process behind the moves sometimes. And can’t play a full game with me yet, but it probably doesn’t know more than some rules so it’s impressive that it tried. Love that it thinks of several candidate moves and sometimes pretty reasonable moves.
@@antonystringfellow5152the more amazing thing is that this is a new small team only a few year together with everyone educated within China, with a shoestring budget. There are dozens of teams like that in China now working on AI. All others don’t want to disclose their research
Its not stealing. Who else built a cutting edge model on a shoe string budget. Also, it uses a mixture of experts model that maximises a small language model. Mixture of experts has been well published, teaching anyone to build it. They may be using llama for data, but that was shared in open source format.
Yeah, David sounds so stubborn when it comes to geopolitics. The whole channel is built on AI euphoria of utopian dreams. Do you really think there will be countries in utopian future? Division on countries is the main thing that slows down the progress. Trillions of dollars, millions of lives are spent on fighting with each other, instead of doing science. How can you try to look in the future with this brainwashed worldview? I don't understand this patriotic bs
Do not underestimate the Chinese, they do have the capacity and if they don't, they will built the capacity they need pretty fast. Just look every other industry, EVs, Solar, Ship building and in the near future semiconductors. In fact the biggest issue with China is that they usually built overcapacity. Different from Soviet Union that was a undercapacity economy.
The capacity is in Nvidia gpus, the top sku is illegal to export to China, they have a limited 'dragon' variant or the -90 series cards You can't build these, you can only buy them, and China can't buy them
Like get hardware from where?. There is only one supplier and it is carefully controlled. Then you come to the researchers, how many good ones do they have?. Maybe 2 or 3, they needed thousands and you can't just "grow" these resources... Nah, China has no chance... 😢
@ Ohh they will built them my friend, no exactly Nvidia GPUs but GPUs in general. ASML CEO given like 10 years to catch up, cut at least 5 of those 10 years, these people are relentless and very academic.
I've been testing R1 on a selection of my approved physics/math problems for Humanity's Last Exam. It did better than o1/o1mini/o1preview on certain problems (it actually solved problems that none of the other LLMs could). The model's really not bad.
As a European, this makes me genuinely optimistic. It gives us hope that we might not fade into irrelevance on the global stage after all. The democratization of AI, driven by open-source initiatives, could empower Europe to close the gap without needing the massive resources of tech giants like the U.S. or China. What excites me even more is the potential for this technology to benefit not just the wealthy, but everyone. If AI truly becomes ubiquitous and accessible, it could level the playing field and bring opportunities to those who’ve traditionally been left behind. This feels like a step toward a future where innovation and prosperity are shared more equally.
I wouldn’t get too excited. You could develop whatever you want but if an American says jump, you WILL go “how high”. Subservience to the US Empire has sealed Europes fate and it will continue on a downward spiral.
China is already ahead of US in many tech sectors like EVs, phones and consumer electronics, internet companies and social media. The only thing I can think of that they are behind is semiconductor and hardware, in which they are not that far behind (like 5 years or so). They will catch up in no time.
@@ay5960 china can make good hardware already but it requires a rewarding market to commercialise it for long run, which is the biggest difficulty for China. Same for operating systems, industry softwares, game engines. the older market is already fully occupied so it's hard for new gamers to take a place even he does qualify. but anyway, china has the power to surpass the US tech, it's just unleashed yet
18:12 you mention North Korea being a villain saying if I can't win then no one can but that literally sounds like the US as well, you guys have this bizarre notion that you have to be number one in everything rather than the logical course of, if we cooperate with everyone then everyone wins
Yeah but the capital class can't have us viewing everyone on earth as equals or else we might feel bad when they are performing slave labor for our corporate overlords.
I think he kinda pointed that out when he said it was the same game theory as Mutually Assured Destruction, which was the entire point of the cold war. Yes, America is unarguably self-centered in almost every way, but that's kind of America's identity. Land of opportunity, where people leave their homeland communities in search of a better life for themselves and their immediate family.
@@charlielarson1350 by going to a place where you have to pay stupid amounts in medical care, risk your children from being shot at school and near enough zero support should you lose your job?
Exactly, the US is the world champion at "we must be #1 or we take the world down with us" thinking. Projecting that mentality onto North Korea is standard deceptive propaganda.
Missed you David, I started with your videos, my man... and now im getting my first AI Degree in march. (and have built so many projects since!) I watched you on Notepad++ with Raven and Huggingface was barley getting started... I learned about Arxiv whitepapers thanks to you and Yannic Kilcher... Thank you so much David... you may never see this but.. the journey to here... such an amazing journey... please dont stop your videos on the edge of new evolutions of what we have already... your vision is so clear.. we need you now more than ever! Thank you for your inspiration David, and I cant wait to see your new content coming! Edit and sidenote: Keep doing what makes you happy! I know you working hard on many things but, we love the Happy D. Shapiro who's STILL leading the change in the world with the rest of us!
To continue with your car analogy: if one car comes with the users manual that allows anyone to fix it and cost 1/20 the cost personal preference seems to be less important.
David, why did you ruin the rest of your interesting thoughts by starting out with such ignorant and racist remarks about the Chinese. Anyone who underestimates the Chinese after the incredible pace of progress they have and are making is a fool. Just take Nature magazine's recent ranking of the best science universities in the world in which China took eight out of the top ten spots. This gives us a glimpse of the quality of China's scientific research. Similarly, you mention the intense competition between US AI companies in which they are obviously learning from each other, but it is only the Chinese that you accuse of stealing ideas!? Last but not least, you say nothing about applying AI advances to the real world which is exactly where China excels. Indeed, China is already applying AI to manufacturing, medicine, education etc. on a much wider scale than the US. I would recommend that you wait a while before writing off China in this technology race.
@@coolcrum98 Actually, I am of Irish origin but living in Brazil. You don't have to be Chinese to object to stupid or prejudiced commentaries against them. Or to respect their hard work and intelligence. We should never forget that it was the Chinese who invented so many things in the last thousand years or more that allowed them to become the largest economy in the world for so long. We talk about them stealing technology yet it was the Europeans who had to go to great lengths to steal the Chinese secrets for making silk, porcelain and gunpowder.
You are born for this work. I’ve picked myself up off the floor multiple times as a direct result of your videos, spurring me onwards. A service to humanity. Thank you!
This is the ChatGPT moment for open-source models. I've tested it on reasoning puzzles and creative writing and it's blowing me away. and I love reading it's thinking or problem-solving process- absolutely fascinating. Was not expecting the quality of creative writings it's putting out. This is the first time I'm choosing to use a free open-source model over paid, closed source models. ClosedAI just got punched in the face.
Really glad you’re back. You’re one of the more forward looking AI enthusiasts I’m following. It makes a big difference that you have other larger philosophical/civilisational ambitions
DeepSeek R1 is like the 9/11 moment - there was a time before and everything changed after it.... let's see how this will unfold in the next weeks. Pace is non linear from now on!
Wow you are so USA-centric. North Korea is isolationist but their actions have been very rational--they think if they have a nuke they will not be bombed and that is absolutely correct. Also when will the US learn to stop underestimating China?
You share so much information that I feel I need to watch your videos several times to "collect" it all. Don't stop doing what you're doing. You are a YUGE contribution!
No compute? No data centers? Isn’t this because U.S. government banned sales to China 😂 Just imagine what the Chinese can do with equal hardware resources. They are certainly making up for it with their brain power 😂
Saying china is stealing research is such a loaded comment without much thought to it. If you are in AI research a lot of works is published and available easily through your university access for the matter of fact. And anyone who has been in ai research knows how big the ai research community is in china. More often than not you see an equal if not more research papers being published from china than from the us or Europe. As the video says everyone knows the high level recipe or the secret sauce of ai. It’s just how you scale it up and that’s largely an engineering and capital issue which doesn’t really need much stealing
I'm glad you took a break. I imagine it was the best thing for you at the time. I'm even more glad you ate back. Hopefully you keep a better work life balance so this is more sustainable for you moving forward. Your voice is important and your videos are enjoyable. I look forward to seeing them, even if they are less frequent than they were before.
Honestly, im not using AI for anything special. But your videos have given me so much hope. I get this warm feeling of optimism thats actually helping me get trough so much stress, and is making me more productive. I generally try to get on with my day like nothings changing, but anytime i get stressed or feel negative i just think ''theres a decent chance this problem is gonna be solved at some point by AI'', and so i calmly go on with my day, ironically, this calm has led to me solving many of my problems myself, when otherwise i might have avoided them or not been as effective at solving them. Love you, brother.
I am really quite glad you found a space to add some dogs to your life. The joy from the relationships we have with dogs especially is extraordinary and the relationship that is built with the environment we live with and the beauty of the season and even daily changes. That commitment is not for every person but it is worth the time we have to share.
I really love the example of the printing press and how by banning it the Ottomans became the sick man of Europe. And this is the broader reason why the west prospered and the east did not from say 14-15th century onwards.
Not really. Western Europeans routinely stole non-Euro tech, used military invasion, destruction of local tech, & dirty tricks like poison, disease, opium to weaken rivals. An aggressive military posture & military tech (guns, steel weapons/armor) & dirty tricks (germs, opium) was the real key. Nonviolent tech was a side benefit.
"You're not gonna starve... I have that much faith in humanity." You mean me? I'm not going to starve? I sure hope you're right, man. But very seriously - How am I different from the millions of people world-wide who struggle today to survive without adequate clean water, nutrition, shelter, protection from violence, and health care? It's not like we can't afford now to divert resources to help those people. The problem appears to be that even personal wealth in the billions of dollars is not enough for those who have been able to amass it. What, exactly is going to get them to change their priorities? The US government has traditionally at least given lip service to address the plight of the "working class". But they haven't spent much effort on the homeless and unemployed. As AGI kicks more and more of us out of the workforce, why will those in power care more about those who are impoverished than they do now?
I’m so happy you have dogs, I’m so happy your health issues are improving, and it is so nice to see you back. You absolutely contribute to the conversation as well as innovative, realistic, and positive ways of framing the impacts of AI in our lives.❤
Hey Dave, glad you are back doing these updates. Just curious, but are we at the stage where these models can be run on a home computer without investing thousands of dollars. Any idea what the basic requirements would be for the average non high tech enthusiasts out here? You know, the kind who might have an older computer and are mildly interested in upgrading if it's not too expensive?
I saw a post yesterday that one of the smaller versions of the R1 model can be run on a mac mini (which costs around 599$) and this version is supposed to be comparable to OpenAI's 4o going by the benchmarks. Didn't try it myself, just thought this might be interesting to you. I think I found the post in David's retweets on Twitter.
I pay for both Claude and Open AI right now because they are better for different things. DeepSeek is really interesting. This stuff is moving so fast, I pity companies who are locking into something too quickly. We should be keeping our options "Open". ahem.
David, you seem like a nice guy, and I enjoy your dissertations, BUT isn't it useful to consider this: AI will accelerate EVERYTHING. Now looking at the last 20 or 30 years, would you say life on earth has evolved in a bad way? I would, more inequality, more anxiety, more irresponsibility, less democracy, more wars and a degradation of tolerance... IF all this is about to accelerate, why hope things for us as a society will get better? I wish I could be optimistic...
If AGI actually pans out that would be a paradigm shift. The world will be fundamentally different after that technology becomes available. In the early days it will be chaos and nobody knows how it will turn out.
The datacenter advantage is not really substantial imo. Chinas industrial capacity is already 25% bigger per ppp gdp than that of the US. So they just have to prioritize data centers to catch up. Also their energy production is far bigger and growing while it flattened more than a decade ago in the US which is the current bottleneck afaik
I love your style of having the bulleted list visible. Few do that, and it makes it much easier to follow, whether I can have sound on or not! -- I also like the colors and style.
Definitely with you on dogs. My partner got them and I was like "I don't have time to look after dogs as well as the kids!".. but they are so interactive and rewarding I changed my mind and now do most of the walking. Also they can really calm your mind; if you're anything like me, trying to keep up with what's happening in AI as well as thinking about what all the implications are can make my head spin like an out-of-control gyro. Keep up the good work!
What kind of job should I be working to get I’m 33 but not sure what fields are safe I want to work in tech but ai just keeps improving so much I’m not sure where to move
I don't really get a lot of new info out of your videos because I'm already really tapped into the AI/Singularity space but it's nice hearing positive news and getting reminded of how its most likely going to play out because a lot of current events have been making me worried more and more that we might tip into AI dystopia, thanks dave!
I will happily accept "Team Mom." 🎯 😁 Gotta keep you here, Dave.
Also mean CUDA isn’t the moat NVDA said it was
As a Brazilian
The prices of Open IA is really prohibitive because the Dollar is too expensive.
DeepSeek will really OPEN the IA for Hundred of countries.
Its a trully Game Changer
The China point is so much more nuanced than what's presented here. Yes there is stealing, mimicry, copying, etc. But look at the surnames on many of the research papers pushing the frontier forward. The Chinese academic system is massive and is churning out unprecedented amounts of bona fide AI researchers and engineers. People who are incredibly smart and have the State apparatus posturing unparalleled resources to back them. They shouldn't be underestimated.
Correct! I saw a research paper the other day. I cant remember exactly what paper it was, but it had 14 researchers on board. 13 Chinese and one middle-eastern. M$ or something..It's the same all over, especially in AI.
100% since they dont have the same kind of regulations they can just build endless coal power plants to produce their power. And let's be real, Asian cultures prioritize education and achievement. If you're great at STEM in the states you're getting bullied.
Congrats on getting new dogs! If my cat doesn't knock something off my desk or beg loudly for treats during work what is even life? Having office animals is the best.
@rawfromnowhere truth.
Isnt open source intended for this type of innovation
All ai companies have at least 40% people of Chinese descent! Many papers have 100%
Look at how US gov restricting Chinese space program which leads them to be more productive and cost effective. Same thing will happen to AI.
Exactly, US only can wish that China is just the modern day Soviet Union.
evolutionary pressure
It just happened with Chinas DeepSeek R1, they beat Open AI version o4 in benchmarks
Looks like a Chinese propaganda account
@@houndsraddforb4284 no.. I am pro-American. but just tell the truth.
thank you china
Bro we need you in this space. Please keep making videos, just do it at your own pace.
I definitely miss him
Word
I agree. David, you have a great ability to cut through the hype and calmly explain the data.
Agreed!
Hey American.... you mean like Europe, with individual Countries under an EU umbrella 😂😂😂
To be honest, DeepSeek R1 being so good caught me completely off guard. But then again, I've learned to accept and embrace surprise in this space.
It's reasoning process is the most sophisticated I've seen to date. And very human like in the way it questions itself.
This company doesn't appear to be behind OpenAI at all. I'd say it's ahead, when taking into consideration the size of the models.
I love the transparency and thinking process of it. It’s fun to try to play chess with it even if it forgets what’s going on because you can see how it thinks. It’s hilarious when it makes illegal moves, but educational to see the thought process behind the moves sometimes. And can’t play a full game with me yet, but it probably doesn’t know more than some rules so it’s impressive that it tried.
Love that it thinks of several candidate moves and sometimes pretty reasonable moves.
@@antonystringfellow5152the more amazing thing is that this is a new small team only a few year together with everyone educated within China, with a shoestring budget. There are dozens of teams like that in China now working on AI. All others don’t want to disclose their research
If OpenAI doesn't open-source, it should rename itself to CloseAI.
This benefits the rest of the world. No more expensive US tech in exchange for political dominance.
Yes the forever cries of stealing, mimicry, copying, etc, all sound sour grapes to me
The way this guy describes 'rogue actors' is literally what the USA is doing.
Rest in Peace OpenAI. You lost everyone’s trust once it turned into ClosedAI.
yes they were distilling from US models, but look at the sheer number of chinese names on papers. they are about to eat the US's lunch.
Its not stealing. Who else built a cutting edge model on a shoe string budget. Also, it uses a mixture of experts model that maximises a small language model. Mixture of experts has been well published, teaching anyone to build it. They may be using llama for data, but that was shared in open source format.
@@krishchetty Typical American response from Dav. When the Chinese build something the America couldn’t they sort to response like stealing etc
If OpenAI doesn't open-source, it should rename itself to CloseAI.
Exactly it was my biggest disappointment!
Times change, everything changes, but David’s opinion of China never changes. 😂
that bigot is so ignorant
Yeah, David sounds so stubborn when it comes to geopolitics. The whole channel is built on AI euphoria of utopian dreams. Do you really think there will be countries in utopian future? Division on countries is the main thing that slows down the progress. Trillions of dollars, millions of lives are spent on fighting with each other, instead of doing science. How can you try to look in the future with this brainwashed worldview? I don't understand this patriotic bs
Do not underestimate the Chinese, they do have the capacity and if they don't, they will built the capacity they need pretty fast. Just look every other industry, EVs, Solar, Ship building and in the near future semiconductors. In fact the biggest issue with China is that they usually built overcapacity. Different from Soviet Union that was a undercapacity economy.
They are first in the majority of critical research fields and first economy by ppp
The capacity is in Nvidia gpus, the top sku is illegal to export to China, they have a limited 'dragon' variant or the -90 series cards
You can't build these, you can only buy them, and China can't buy them
Like get hardware from where?. There is only one supplier and it is carefully controlled. Then you come to the researchers, how many good ones do they have?. Maybe 2 or 3, they needed thousands and you can't just "grow" these resources... Nah, China has no chance... 😢
@ Ohh they will built them my friend, no exactly Nvidia GPUs but GPUs in general.
ASML CEO given like 10 years to catch up, cut at least 5 of those 10 years, these people are relentless and very academic.
@@matt5721 they buy it through setting up companies in Malaysia ,HK , Singapore etc.
I've been testing R1 on a selection of my approved physics/math problems for Humanity's Last Exam. It did better than o1/o1mini/o1preview on certain problems (it actually solved problems that none of the other LLMs could). The model's really not bad.
A country without freedom never can creat any important,we should not worry about it
The R1 model has been a lot better than GPT for deeper thinking based on my testing.
@@qingbomo7160are you serious or try to be joking?
Not bad? You mean impressed?
@ yep
As a European, this makes me genuinely optimistic. It gives us hope that we might not fade into irrelevance on the global stage after all. The democratization of AI, driven by open-source initiatives, could empower Europe to close the gap without needing the massive resources of tech giants like the U.S. or China.
What excites me even more is the potential for this technology to benefit not just the wealthy, but everyone. If AI truly becomes ubiquitous and accessible, it could level the playing field and bring opportunities to those who’ve traditionally been left behind. This feels like a step toward a future where innovation and prosperity are shared more equally.
Europe will continue to fall behind until it changes its ways.
To be fair Europe invented much of this technology. Just didn’t commercialise it. But in the academic sphere we are at the forefront.
I wouldn’t get too excited. You could develop whatever you want but if an American says jump, you WILL go “how high”.
Subservience to the US Empire has sealed Europes fate and it will continue on a downward spiral.
@TheMinimumPC cool story bro
How is Europe even relevant?
China is already ahead of US in many tech sectors like EVs, phones and consumer electronics, internet companies and social media. The only thing I can think of that they are behind is semiconductor and hardware, in which they are not that far behind (like 5 years or so). They will catch up in no time.
@@ay5960 china can make good hardware already but it requires a rewarding market to commercialise it for long run, which is the biggest difficulty for China. Same for operating systems, industry softwares, game engines. the older market is already fully occupied so it's hard for new gamers to take a place even he does qualify. but anyway, china has the power to surpass the US tech, it's just unleashed yet
18:12 you mention North Korea being a villain saying if I can't win then no one can but that literally sounds like the US as well, you guys have this bizarre notion that you have to be number one in everything rather than the logical course of, if we cooperate with everyone then everyone wins
Yeah but the capital class can't have us viewing everyone on earth as equals or else we might feel bad when they are performing slave labor for our corporate overlords.
I think he kinda pointed that out when he said it was the same game theory as Mutually Assured Destruction, which was the entire point of the cold war. Yes, America is unarguably self-centered in almost every way, but that's kind of America's identity. Land of opportunity, where people leave their homeland communities in search of a better life for themselves and their immediate family.
@@charlielarson1350 by going to a place where you have to pay stupid amounts in medical care, risk your children from being shot at school and near enough zero support should you lose your job?
@@TheBoringInvestorMan yes
Exactly, the US is the world champion at "we must be #1 or we take the world down with us" thinking. Projecting that mentality onto North Korea is standard deceptive propaganda.
Missed you David, I started with your videos, my man... and now im getting my first AI Degree in march. (and have built so many projects since!) I watched you on Notepad++ with Raven and Huggingface was barley getting started... I learned about Arxiv whitepapers thanks to you and Yannic Kilcher... Thank you so much David... you may never see this but.. the journey to here... such an amazing journey... please dont stop your videos on the edge of new evolutions of what we have already... your vision is so clear.. we need you now more than ever! Thank you for your inspiration David, and I cant wait to see your new content coming!
Edit and sidenote: Keep doing what makes you happy! I know you working hard on many things but, we love the Happy D. Shapiro who's STILL leading the change in the world with the rest of us!
China doesn’t have the resources for tech development? You are kidding yourself there!
My favorite part is where your dogs volunteered to participate in your advanced super-alignment demonstration: "Hey, be nice!"
I cannot overstate the importance of your videos.
Is this a real comment?
@@Lugmillord glazing
To continue with your car analogy: if one car comes with the users manual that allows anyone to fix it and cost 1/20 the cost personal preference seems to be less important.
David, why did you ruin the rest of your interesting thoughts by starting out with such ignorant and racist remarks about the Chinese. Anyone who underestimates the Chinese after the incredible pace of progress they have and are making is a fool. Just take Nature magazine's recent ranking of the best science universities in the world in which China took eight out of the top ten spots. This gives us a glimpse of the quality of China's scientific research.
Similarly, you mention the intense competition between US AI companies in which they are obviously learning from each other, but it is only the Chinese that you accuse of stealing ideas!?
Last but not least, you say nothing about applying AI advances to the real world which is exactly where China excels. Indeed, China is already applying AI to manufacturing, medicine, education etc. on a much wider scale than the US.
I would recommend that you wait a while before writing off China in this technology race.
@@coolcrum98 pink hand typed this..
@@coolcrum98 pigs knuckle?
@@coolcrum98 Actually, I am of Irish origin but living in Brazil. You don't have to be Chinese to object to stupid or prejudiced commentaries against them. Or to respect their hard work and intelligence. We should never forget that it was the Chinese who invented so many things in the last thousand years or more that allowed them to become the largest economy in the world for so long. We talk about them stealing technology yet it was the Europeans who had to go to great lengths to steal the Chinese secrets for making silk, porcelain and gunpowder.
@@coolcrum98 reported
@@martymarl4602 oh no!!!!!
You are born for this work. I’ve picked myself up off the floor multiple times as a direct result of your videos, spurring me onwards. A service to humanity. Thank you!
This is the ChatGPT moment for open-source models.
I've tested it on reasoning puzzles and creative writing and it's blowing me away. and I love reading it's thinking or problem-solving process- absolutely fascinating.
Was not expecting the quality of creative writings it's putting out.
This is the first time I'm choosing to use a free open-source model over paid, closed source models.
ClosedAI just got punched in the face.
Thank you so much for coming back! Your videos bring me a sense of peace about the future. I'm really grateful!
Let's hope that peace is real ;)
Glazing
@@tophatv2902not even slightly. Maybe I haven't found any other youtubers who cover this stuff like this, but I am being honest.
Happy you’re back! Your commentary really adds a lot of value and is super helpful and insightful.Thank you!
Glad to have you appearing more often - you have the gift of insight that very few other content creators have. I Thank you !!!
Really glad you’re back. You’re one of the more forward looking AI enthusiasts I’m following. It makes a big difference that you have other larger philosophical/civilisational ambitions
The interruptions are barely noticeable to the viewer. FYI
Agreed! Tbh him acknowledging the "interruptions" was more distracting than the dogs themselves 🤣
So good that you are back.
Love these videos!
I still really like this style too - I really enjoy watching every minute of them.
DeepSeek R1 is like the 9/11 moment - there was a time before and everything changed after it.... let's see how this will unfold in the next weeks. Pace is non linear from now on!
Don't get a dog, get a cat.
We all need to get used to living with a potentially dangerous entity that thinks very little of us.
Wow you are so USA-centric. North Korea is isolationist but their actions have been very rational--they think if they have a nuke they will not be bombed and that is absolutely correct. Also when will the US learn to stop underestimating China?
The Ottoman Empire no longer exists because it relied on Byzantine Generals. Vernum Cypher... It's all about the namespace frequency.
You share so much information that I feel I need to watch your videos several times to "collect" it all. Don't stop doing what you're doing. You are a YUGE contribution!
No compute? No data centers? Isn’t this because U.S. government banned sales to China 😂 Just imagine what the Chinese can do with equal hardware resources. They are certainly making up for it with their brain power 😂
Sourgraping?
Thank you Dave! Give your dogs a hug, take one yourself. Peace to you
Love the way David decomposes complex and nuanced subject matter into individually addressable components. It's like unstirring the cream!
Welcome back. Missed your quality AI content.
"There is no moat," gets a mote in the eye.
I appreciate the content!
Saying china is stealing research is such a loaded comment without much thought to it. If you are in AI research a lot of works is published and available easily through your university access for the matter of fact. And anyone who has been in ai research knows how big the ai research community is in china. More often than not you see an equal if not more research papers being published from china than from the us or Europe. As the video says everyone knows the high level recipe or the secret sauce of ai. It’s just how you scale it up and that’s largely an engineering and capital issue which doesn’t really need much stealing
I'm glad you took a break. I imagine it was the best thing for you at the time.
I'm even more glad you ate back. Hopefully you keep a better work life balance so this is more sustainable for you moving forward. Your voice is important and your videos are enjoyable. I look forward to seeing them, even if they are less frequent than they were before.
Honestly, im not using AI for anything special. But your videos have given me so much hope. I get this warm feeling of optimism thats actually helping me get trough so much stress, and is making me more productive. I generally try to get on with my day like nothings changing, but anytime i get stressed or feel negative i just think ''theres a decent chance this problem is gonna be solved at some point by AI'', and so i calmly go on with my day, ironically, this calm has led to me solving many of my problems myself, when otherwise i might have avoided them or not been as effective at solving them.
Love you, brother.
I am really quite glad you found a space to add some dogs to your life. The joy from the relationships we have with dogs especially is extraordinary and the relationship that is built with the environment we live with and the beauty of the season and even daily changes.
That commitment is not for every person but it is worth the time we have to share.
I really love the example of the printing press and how by banning it the Ottomans became the sick man of Europe. And this is the broader reason why the west prospered and the east did not from say 14-15th century onwards.
Not really. Western Europeans routinely stole non-Euro tech, used military invasion, destruction of local tech, & dirty tricks like poison, disease, opium to weaken rivals. An aggressive military posture & military tech (guns, steel weapons/armor) & dirty tricks (germs, opium) was the real key. Nonviolent tech was a side benefit.
So whats the new valuation of OpenAI?
"You're not gonna starve... I have that much faith in humanity."
You mean me? I'm not going to starve?
I sure hope you're right, man.
But very seriously - How am I different from the millions of people world-wide who struggle today to survive without adequate clean water, nutrition, shelter, protection from violence, and health care?
It's not like we can't afford now to divert resources to help those people. The problem appears to be that even personal wealth in the billions of dollars is not enough for those who have been able to amass it. What, exactly is going to get them to change their priorities? The US government has traditionally at least given lip service to address the plight of the "working class". But they haven't spent much effort on the homeless and unemployed. As AGI kicks more and more of us out of the workforce, why will those in power care more about those who are impoverished than they do now?
So super intelligence by the end of this year or the next? Hard takeoff style.
Big fan! Great you are back! Based on recent discoveries, would you update the video with your predictions? Thanks!
Thanks for continuing these, they are excellent and helpful
We love having you back! Your voice is definitely a huge contribution to the field!
Openai preipo market cap dropped by half last week.
Now the real competition begins. Can Sammy get that golden egg back?
I'm glad to see you putting out videos again. I so much enjoyed them and missed them
I’m so happy you have dogs, I’m so happy your health issues are improving, and it is so nice to see you back. You absolutely contribute to the conversation as well as innovative, realistic, and positive ways of framing the impacts of AI in our lives.❤
Welcome back, and thanks to Julia for her encouragement.
Thank Julia for bringing our champion back
Glad to see you back! Keep it flowing
Love your points on cognitive hyper abundance. Ultra exciting!! Loved this whole presentation. Fantastic job!!
I am blown away. Running 14B version on standard mac mini using LM Studio. It seems to me on the same level of OpenAI for my use cases, wow
Extremely enlightening. Thank you Dave :)
Hey Dave, glad you are back doing these updates.
Just curious, but are we at the stage where these models can be run on a home computer without investing thousands of dollars.
Any idea what the basic requirements would be for the average non high tech enthusiasts out here?
You know, the kind who might have an older computer and are mildly interested in upgrading if it's not too expensive?
I saw a post yesterday that one of the smaller versions of the R1 model can be run on a mac mini (which costs around 599$) and this version is supposed to be comparable to OpenAI's 4o going by the benchmarks. Didn't try it myself, just thought this might be interesting to you. I think I found the post in David's retweets on Twitter.
@missoats8731 thanks, I appreciate that 🙂
Welcome back! Thank you for the insightful content.
Exciting that you're back Dave, there really is no one that does it quite like you.
So happy to see one of these videos again, Dave. And more dogs please. 😉
didn't you say you were quitting this channel or did I multiverse shift again? glad you're here btw.
Really glad you listened to Julia.
So glad to have you back!
I pay for both Claude and Open AI right now because they are better for different things. DeepSeek is really interesting. This stuff is moving so fast, I pity companies who are locking into something too quickly. We should be keeping our options "Open". ahem.
David, you seem like a nice guy, and I enjoy your dissertations, BUT isn't it useful to consider this: AI will accelerate EVERYTHING. Now looking at the last 20 or 30 years, would you say life on earth has evolved in a bad way? I would, more inequality, more anxiety, more irresponsibility, less democracy, more wars and a degradation of tolerance... IF all this is about to accelerate, why hope things for us as a society will get better? I wish I could be optimistic...
If AGI actually pans out that would be a paradigm shift. The world will be fundamentally different after that technology becomes available. In the early days it will be chaos and nobody knows how it will turn out.
The datacenter advantage is not really substantial imo. Chinas industrial capacity is already 25% bigger per ppp gdp than that of the US. So they just have to prioritize data centers to catch up. Also their energy production is far bigger and growing while it flattened more than a decade ago in the US which is the current bottleneck afaik
welcome back davieboy! we missed ya
I’m glad to see you on here. I thought you were leaving. I’m glad you didn’t. Hooray you’re one of my favorite channels.
Totally. After several hours of using it for math, I think it's probably closer to o3 benchmarks. I am floored.
What can you use it for and what shouldn't you use it for
I love your style of having the bulleted list visible. Few do that, and it makes it much easier to follow, whether I can have sound on or not! -- I also like the colors and style.
Nice to see you back and sharing your thoughts
Honestly, if they are going to "steal" it, launder it and release as Apache 2.0; I really don't mind.
"The water's warm, come on in."
So happy you are back, thanks for your evaluation!
Definitely with you on dogs. My partner got them and I was like "I don't have time to look after dogs as well as the kids!".. but they are so interactive and rewarding I changed my mind and now do most of the walking. Also they can really calm your mind; if you're anything like me, trying to keep up with what's happening in AI as well as thinking about what all the implications are can make my head spin like an out-of-control gyro. Keep up the good work!
What kind of job should I be working to get I’m 33 but not sure what fields are safe I want to work in tech but ai just keeps improving so much I’m not sure where to move
Nice to have you back
Hey David, how long until we can let go?
Keep bringing the material. You explain it well.
Happy to see you my friend :)
I'm already using a Chinese AI that converts photos to 3D models. It's free and can run locally. ❤
"One model to rule them all" just in case we are paranoid.
Really appreciate you, thank you for all you do!!!
Prodigal Son is back😊
So glad to have you back. The only voice I truly trust in this space. The "no hype, so stick it in your pipe" coverage I want. ❤
We missed you brother… we seriously need people like you
Davis! you’re a good human. Your signal matters, and we need it. Keep radiating buddy.
Davis!
Stolen? With all the sanctions?
This makes stargate more interesting/clear- OpenAI advantage shifting to having contractual rights to superior resources than competition
As the curve gets steeper timelines are compressed.
I don't really get a lot of new info out of your videos because I'm already really tapped into the AI/Singularity space but it's nice hearing positive news and getting reminded of how its most likely going to play out because a lot of current events have been making me worried more and more that we might tip into AI dystopia, thanks dave!