"Greatest lie ever"? not really. This channel is just stating the obvious. Of course the announcement of DeepSeek R1 was meant to disrupt the market led by US tech companies promoting AI. Those were overvalued anyway and the DeepSeek announcement provided the market correction that we need. This channel also failed to mention that DeepSeek is open source and their methodologies are published unlike OpenAI, despite its name, is not open at all. If you are running DeepSeek through their app or cloud service, yes it is censored by China and can be a security risk. However, you can run their model locally even without internet connection or without connecting to their service so in this case security is not an issue. Also because it is open source, anyone can look at it on how the technology works and be able to replicate it. This is a good thing since the technology is no longer monopolized by big tech companies.
I'm running Deepseek locally and it is censored. Ask it about the Tianman Square protests and massacre. You'll see. By running it locally you simply avoid giving your data to China.
Thats is what they are trying to ruin... Cant ruin What is Free to the public... Destroying the Monopoly of Ai training is what they are pissed at.. but they forgot that Human with limitation exceed more than a bottomless pit of resources.
Heres the thing Its not even Good, it makes more mistakes than GPT 3.5, Fake TH-cam videos showing o1 takes over a min to respond ? I use o1 daily it takes almost no time before answering (1 second or less) Not over a min
This video had very little information about how DeepSeek is a lie. This video barely addresses possible lies about the cost of training and the issue of whether R1 is a distilled model using data from OpenAI. This video is an okay overview of the last week of news around DeepSeek, but the topic in the title is barely addressed.
It's open source you can easily ask any person who's proficient in machine learning if it has any resemblance to openai. Also it's arrogant to think that china hasn't been working on any sort of ai model for the past decade.
@@somejimmydude2121 seeing how Zuckerberg and musk have both given up on screening posts, the safety rails of AI will also probably be withdrawn soon enough.
Honestly, Moon could've done better with this one, the AI sponsor doesn't help his credibility either. A bit more unbiased research, and he could've painted a better picture that reflects the AI race. I know people in the west won't accept it but the release of Deepseek is arguably the best thing that's happened to the industry, an open source AI produced by reinforcement learning that's on the same level or even better than OpenAI's chatgpt O1 model. Ironicaly Silicon Valley will benefit immensely from this coz it's open-sourced, so they can literally see process of how the R1 model "thinks", and perhaps apply the lessons in their own models. It's true, the cost of the R1 model is up for debate, but still even if the real cost is 10X or 20X the 6 million dollar cost, it's still way more cheaper than all the American AIs. It's crazy to think that China is giving the world a real open AI, so developers across the world can alter the model for their own use, and it's on the par with the best Chatgpt model. Most importantly, the entire world can it for free, in contrast to OpenAI's O1 costing hundreds of dollars. Taking into account, the geopolitical political tension between the 2 superpowers, data privacy concern is a real issue. Another beauty of the Deepseek R1 model is that you can run it locally, which may keep all your data on your system. I'm glad Deepseek took the giant leap forward to wake up Big Tech. Google, Microsoft, OpenAI, Apple, all these companies' main focus is gatekeeping the AI industry, racing against each other for a monopoly grip on this game changing technology, crushing all start-ups that could challenge them.
Dude, this channel can't even get basic general history right (cf. the video on how USA ruined itself or something like that). Seriously, some basic stuff about The Enlightenment that here in Europe is tought in high school and this channel can't even get that right. Truth is that this is one of those channels that appears smart only to stupid people.
damn lemme just try out that Deepseek and see if its better than chatGPT i didnt know there was a better AI you can just download right now ty for the info lol
I couldn't care less about Sam and his crying over his ChatGPT data being stolen. He stole them from everyone else and monetizing it, he reaped what he sow tbh. Open-source is good, it always support open competition
damn lemme just try out that Deepseek and see if its better than chatGPT i didnt know there was a better AI you can just download right now ty for the info lol
It's not China's fault that the stock market dropped after the release of DeepSeek. The real issue is that many US corporations have invested excessively in artificial intelligence without achieving significant results. As a result, the stocks of several US companies are overvalued, lacking genuine improvements to justify their prices. Additionally, Western regulators have imposed strict controls on the AI industry for several years, hindering its development, as has occurred with the social media sector.
DeepSeek got released an entire month BEFORE the crash. This was market manipulation, helped by a few well timed hit pieces. The entire premise of this video is invalid. As per usual.
@@lxpyes u can, altho to run the original version you would need a large amount of vram, u can still get the smaller models with more manageable ram requirements altho they won't be as powerful as the original
So as an average person can anyone tell me why I should be concerned about China having my data? It is more worrying for my own government, companies in my own country have my data; they can use it against me. Unsure what the Chinese government will do with any of my data. Sorry if I come across a little naïve.
Journalist : how did they got all these cheaps ? Elon Must i sold it to them ... obviously they have the cheap i even cut down the price in half Journalist : ...
Kai-Fu Lee has been working on AI development in China for over 10 years. He worked for Apple and Google respectively before starting his own AI consulting business. The emergence of DeepSeek shouldn’t have been a surprise to anyone, if you know what China has been doing in the last 20 years.
Alexander Wang - the guy in the interview is a bad mount guy. what he said is his elaboration. he even dont need to show any investigation backing for his point. AI roots for the last 20 years as you said under many names.
You mean stealing from people and companies that actually did the work as they are devoided of any capabilities of innovation and "outside the box" thinking no matter how hard they try? Yeah, effed China. They are solely good at one thing; stealing.
You mean he spied at Apple and Google just like the majority of Chinese tech does to America and what they don't get through spying they outright steal a version of it and send it back.
I feel like it will me more often like that, i saw this already on his video about tik tok. He trys to put mor and more his own agenda in his videos i feel like. Sad..
@@Le_caylemoneyeah I'm starting to noticed he's like one of those lite right wing type channels... not super maga but leaning toward that side and pushing it while trying to not look like it
And another thing, deepseek said they used less hardware than OpenAI, not believed, a nerd in the US said it believes they have more GPU's without evidence, instantly believed. Americans just can't stand the fact that they aren't at the helm of everything tech.
The greatest lie Told was that we need billions of Dollars to create powerfull AI. When a hedge fund created the same thing with less than 1% of that Budget.
I would say it's more akin to the US being the dominant empire in the world and doing what it can to stop another empire from growing across the planet. Hmm, maybe that is akin to a Cold War. This is very new to human history there is no record of two great empires existing at the same time like this, where one can so greatly influence the other with something as simple as an application on your phone. No Rome vs Greece. or the Persia vs England. One slip-up could change the face of the world. and Unfortunately for the United States our leadership has never been more questionable while China's technology has never been better. I'm getting excited.
If you saw the Chinese military you would laugh yourself to death. They are so amateur compared to the us. They do have great nerds though. Slave nerds but smart nerds nonetheless.
It's about being too lazy and/or stupid to think beyond such a simplistic explanation. Go trade some GME and AMC, while more adept people do the thinking.
This is a propaganda channel, it's owned by a larger company I think based in Israel and/or France - there's hundreds like this that pump out content pretending to be individual/authentic youtubers.
@numerum_bestia entomology the study of language Shows the word religion literally means To hold back and to bind We don't need middlemen to have a connection to source And religion is the middleman
Not saying it's the case here, but China does have a spectacular track record of IP theft. One should be wary of anything that comes out of an authoritarian country, especially one whose long term results often don't live up to expectations. That said, tech stocks here were a giant circle jerk of constant rising without the evidence to justify it, and a correction was inevitable. Competition is a wonderful thing, and now everyone has a new baseline to work from moving forward.
Lol. Americans. Man, Deepseek is OPEN SOURCE. Your can run it locally in your own machine, host it yourself. No need for their servers. That's the beauty of it. The Deepseek 32B model performs much better than GPT4.
Uh, genus, CENSORSHIP??? Hell, ask Deepseek about the Philippine Sea and they marked as China's claim. Or how about giving the negatives to Mao's regime?
@@finncapistrano9879 What are you talking about? I'm running the Deepseek model locally and there is no censorship whatsoever. I have ran the two prompts you mentioned were "heavily censored" and have gotten basically the same answer for both. I'm absolutely willing to give you both prompts + outputs from ChatGPT's 4o Model & Deepseek R1's 30B Model so you can compare the output via pastebin or something :) But i'm pretty sure you have no interest in that as you only want to propagate that the release of an open source model comparable to the 4o Model is a bad thing.
by making it open source, deepseek gave everyone the option to host it themselves, or even other companies to host it and offer it to users. that should completely alleviate the privacy/security concerns. unless you use the webUI or the app of deepseek directly. then your data goes to china. be clever about it! regarding geopolitics: again the big message is that its open source and the methods of training are revealed. that levels the playing field quite significantly in terms of ai world domination, because it gives everyone a new start line to work off of. and in the current state of affairs, this benefits everyone on earth except the US. regarding nvidia: the training cost of deepseek was way lower than a comparable US model. but that doesnt mean you need less GPUs, it just means you get farther with the same. the stock dip of nvidia was an unfounded investor panic attack. microsoft stock dip was justified, as everyone believed "open"AI was further ahead than they actually are.
First reasonable take ive seen here. If they made deepseek closed and online only, I wouldnt buy it for a bit. but I ran the 32b model and its straight up better than gpt4. seeing how insanely cheap it is to have the full model hosted in america like 2$/1m tokens there is literally no reason to pay openai for sh@t now. if it was all nonsense western AI would have ignored it, like the many other mid models out there. but tbf the nvidia drop was from two angles, the AI stuff, and the 5000 series being an unmitigated disaster, worse performance than 4 series in some cases, worst generational "uplift" ever?, highest relative price increase (ever??), and worst availability.... EVER????? its actually insane how bad its going for them.
@@mrrooter601 there's a disinformation campagin going on performed by US operatives, OpenAI and china haters making this all about China vs US- but this is actually Open source vs Closed source- and they are banking on common folk not reading the deepseek paper or having technical knowledge to understand the difference
@mrrooter601 good addition, i forgot to mention that inference is also way cheaper with ds. agree with the 5000series being crap, but gaming is an afterthought for nvidia and investors, so no real influence on the stock price imo
@ nah nvidia is literally one of the most "normie" stocks ever, combine the normies seeing the 5 series being awful, and seeing a selloff and its compounding. hell the stock started tanking just after the card announcement. now we just started getting the real numbers yesterday for performance too, but its impossible to tell now. Nvidia may not be a gaming company anymore but that doesnt stop idiots for investing in it like it is.
Deepseek didn't come out of nowhere. The v2 and v3 models were also great and comparable to Open Ai's 4o models. The reason why deepseek is this popular has a lot to do with the reasoning model being so much more efficient and distilled models of R1 that can be run very well on consumer desktops. It doesn't require the Nvidia CUDA architecture. There's a lot more to do with it, but the code efficiency increase is downright phenomenal.
It is open source, so you can build your own custom agants. The techniques used are smart, canny in the face of US sanctions. it has already been superceded, so not to worry !
Ali Baba released Qwen 2.5 , beating DeepSeek’s benchmarks. There are probably lots of up-and-coming models that have been working hard in the background.
Also I love how efficient China is at delivering cheap, yet also high quality work. Most things we own are manufactured by China either way, so even if governments try to control the hold they have over our economy, it'll just be a big fat waste of time
So basically the whole videos comes down to, they got to be lying about the price. But then again I don't understand how you can say their lying about the computing power when you can download a local version of deepseek. Totally isolated from Chinese servers. So maybe it was made with the hypothetical 50,000 gpus but clearly they are not need to maintain it.
once the model is train, you dont need to keep training it, and you needed at least a billion to train is the point of this video, so no, deep seek to get to the level we have, didnt cost only less than 10 millions... thats the lie.
@charlestrudel8308 The video does not outright said state 1 billion dollars only stating same guy guesses that it would take 50k gpus to make deepseek then Moon again estimates that 50k of those gpus would cost over 1 billion dollars alone not counting the cost of man hours. Obviously they probably spent over 6 mil. But this video is all just guesses n estimations. Perhaps what's more scary than lying about the price is the idea that the actually found a more efficient way to make AIs.
@@xlegacyzero8878 even if the guess is valid, its still like 25% of the price of the other AI trainning going on, its still quite cheap. and also, a part of the billion i mention is the overhead company expense that are ignored in the initial 6 millions.
@@charlestrudel8308 do you think when we talk about the almost $1 billion dollars to train GPT-o1 we're including the cost of GPUs or salaries??? The cost is measured one-for-one for JUST TRAINING. If we do the same analysis then o1 cost probably tens of billions if not even more!! Please think harder
Space Race: Let's see which superpower nation can create enough force to break past gravity and reach beyond the atmosphere AI Race: Let's see which superpower nation can generate more words
@@Joe-f1k2c Correct, it's about generating huge, animated catgirl boobies. At the moment USA has the dominant market on animated catgirl boobies but China wants some of the action
Its open source - thats the whole thing, you can instead of making a fake title like this, copy the code and DO WHATEVER YOU WANT with it....thats like deepseeks all...."greatest lie ever told" haha
I like how people believed China's claim at face value. This is the same country where they had a "natural" waterfall at a park, until someone went over the waterfall, only to see a tube spewing water, and post it on the internet.
or when they said covid couldn't ever come out from the virology lab that is in the city where the breakouts started and it came from bats, out of all animals that have a lot more contact with humans. Bats...
I like how people automatically believe the west and dismiss everything from China. If you read DeepSeek’s paper, you would know DeepSeek wasn’t built with only 6 million dollars.
People also believe any American claim at face value despite the US running the largest Ponzi scheme in history called “social security”. People will constantly believe anything at face value despite a history of lies. Just look at influencer culture.
Played it for background noise but sounded like his usual BS, asks a few questions why, barely or vaguely answers the questions and only just covers the basics with a bunch of filler information. This dudes videos are all the same, seen one seen them all. I get way more context reading the comments
I doubt China did it so cheaply, but I also doubt that U.S. tech firms are going to do anything other than claim they need the most investment dollars possible. At least some of what our tech industry is doing is price gouging for personal gain.
@@clintholmes2061 He's low information and low critical thought like 80% of the rest of the viewers of these channels who haven't read a book or a few articles of research to save their life.
Lying about the price doesn't matter. cause the mathematics of it still show that somehow their still about to Chatgpts 4 levels without the $200 price tag and they are somehow still in the green. Possibly proving that American companies have overlooked something vital. it's like if the US invented Helicopers while China invented cars for the average citizens, Why would any person buy a helicoper when the car is 20x cheaper.
Lol Americans US feel the NEED to ban the export of "high-end" chips to China to slow down it's technological advancements, but China did it anyways. I'm loving China
Apart from the points you made, Deepseek is the perfect tool for social engineering and info collection, compared to profit oriented open ai, deepseek doesn't need to generate profit. China will shoulder it anyway.
bro, its open source. you can run the distilled models on a literal smartphone locally.the 32b model runs on 32 gb of ram and a modern gpu, and its on par with gpt4. china cant collect data if you run it locally. china cant collect data if you host it on american servers (if you want the full model), its literally free to download. obviously if you use their app they are going to collect data on your prompts and use them for further training, thats literally how it works for every AI.
bruh for god's sake its open source that you can run on your own and literally all "free" AIs on the market exist to collect, sell and train on data average third grader and their mom will sure as hell use "free" Deepseek over paid ChatGPT 4
@ good news, they explained how they did it. now american startups that want to can follow the open source paper and make their own. I mean, if you want to, go run the smaller versions of "deepseek" which are actually llama (large language model meta ai aka. facebook) trained on deepseeks outputs. unless you would only be happy with something that had no chinese involved at all. but lets be real thats pretty uhh ist. Its not even biased towards the CCPs agenda nearly as much as openais gpt pushes DEI. sure you ask about Tiananmen Square on the app and it will get denied, cause that goes to china directly. and they filter messages going to and from the app. but asking the (full size) model self hosted, with 100% ownership here in america, and it will give it to you straight. good datasets are hard to find, and since they were actually trying to make the best AI they could, they mostly avoided poisoning it with shitty CCP propaganda, hence the external filtering. Advancements in AI are going to be a global thing now. since they graciously destroyed "open"AIs grift with the power of open source. Obviously the chinese government has done many things wrong. even if they were even significantly involved here, this they did nothing wrong other than making a better product, for less, and causing a government funded monopoly to seethe. They PROVED state of the art AI can be 100x more efficient, and that is likely one of the best things the chinese have ever done for the planet. so perhaps we should be grateful they did so and shared how to do it too ya know? maybe give them some credit instead of being a bit ist...
I'm curious how much more effective or less effective it is to western AI in social engineering Chinese propaganda is so transparent in general. Although western propaganda has gotten pretty bad itself. I think the more aware of social engineering the general public becomes the less effective it is. Although if one comes up with a new idea for pushing it the learning curve will ensure many fall for it before word can get around. Although how many new methods of pushing the same establishment goals can they come up with? Imo there's only so many ways to convince people to blindly support the establishment.
US is losing and will keep losing. While US were focusing on wars and oil and Middle East and terrorism that most Americans actually think was a inside job, China was focused on technology and finance and global expansion. I think China has shown the world it is right now perhaps the most stable and reasonable country in the world and I say this in a context where the US and Europe are only focused on wars and China had the perfect opportunity to go ape like everyone else right now and invade Taiwan. But they didn't. They have been working for peace while channeling their energy to solve issues like these - a decaying superpower that is crumbling and wants to prevent them to get chips. I wonder if persons realise how insanely dumb that is. In the world we live in there will always be someone ready to smuggle those marvelous chips everyone is talking about right now the same way back in the 80's there was a nut job of a Russian pilot willing to steal a MIG mid training and fly all the way to Hokkaido risking his life because the fuel barely made it and when he got there he didn't know where and how to land. But the US got their hands on that MIG didn't they?
As soon as I saw the deepseek release, I immediately assumed all of this even though I knew none of the specifics. I’ve been following US/China relations for years but recently through the perspectives and narratives from natives and people who lived there and know it without the government censorship. The biggest problem with AI is the mindset controlling it because once a superior model emerges, it’s up to the humans who are utilizing it to decide what its morality will hinge on. Nobody wants China to be in that position.
Stop twisting the truth. I've got DeepSeek and it runs flawlessly on my old GTX, the model is open-source so you can test it yourself. You do realise that, ultimately, it was the banks/hedge-funds etc. that cashed in the biggest by shorting tech stocks. They just wait for industry-disruptions to make big moves, that's the game.
Guys remember when the American rich spoke of greatness of the free market (and now block the export of nvidia chips with sweaty brow)... how capitalism breeds innovation (and see all the planned obsolence and dlc:fication of purchased products) and other funny memes... These guys speak so much bs that it's probably hard to pull occasionally a decent breath in.
The reason it used so much less resources was just stealing from openai. But once theres no one to steal from the resources needed will become the same. The good thing is they released an open source version, which should become an infistry standard
This channel is good at peddling propaganda but usually it's more subtle. Still don't know if it's deliberate or just some dumb kid who exposed himself to a specific kind of content too much. Hope the latter, but the more I watch from this channel (usually avoid it like the plague) the more I'm scared it's the former.
The more competition on the Ai market, the better. $200 a month is ludicrous. No one company should have a monopoly. Just look at the good that did Nvidia. They made a killing and a half.
The name of the game is optimazation more so than power. If things keep going the way they are we will be able to use an ultra powerful ai on a dinky phone!
People in the comments acting like they knew all along 😅 And then you look through their history to find that isn't the case. In almost every single comment lmao.
The fact that black rock lost something, nothing short of amazing. The fact the 800s are just 100s programmed to run not as good, nvidea could face a huge investigation. No smoke without fire. Oooof
@@WillmobilePlus It's not even about CCP bots attacking content creators. We're saying this out of love for the channel. I've been watching "Moon" for years, but I've never been so disgusted by the bias in this video. I'm fortunate enough to being familiar to the AI race to AGI, and Deepseek in particular. Moon has always been one of those channels that are a breath of fresh air, and great analysis of the status quo. But this is most biased video I've watched in this channel, without any intention of giving the audience a clear reflection of the truth. I'm so disappointed, and unsure if this was deliberate or not but Moon dropped the ball on this one.
@@loyisomasango5149 Yeah, sure. Spare us the playact. He was critical of some thing out of China, and like on other videos of the same, here you guys come with just variations of the same BS post about how "he dropped the ball" and "I'm so disappointed blah blah blah". The usual crap Chinese bot farm posters type.
Uhhh this is not a good argument as we can all see the open source model and it can quite literally be trained on simple h20s for 1/1000th the cost. Like you dont even need h100s or h200s!!
another factor is that deepseek is able to run so lean because it focuses on the highest quality data for training Data, as with almost all fields, 80% of the value comes from 20% of the set. Deepseek only uses the most valuable 20%, and can therefore run light and train fast. But how does it know what data is valuable? Easy, it steals and cherry picks it. Short term you look brilliant, long term you will hit a wall pretty quick
This video is a classic case of someone trying to milk a topic by presenting an alternative viewpoint just to create drama and get some of the attention. To sum this up: *'Yo guys. This new thing that is trending rn because it's functional, free, was cheap to create, is open source etc... is... bad.. actually.. somehow... Idk how, but lemme just pretend like it is....* *Btw, let me introduce the sponsor of another AI company for this video...'*
this is just like a research paper. It wouldnt cost much effort if there are already studies supporting yours. Especially when its the same exact topic.
This is the first video of Moon I'm giving thumbs down. Focus shouldnt be on the cost of deepseek, but on Open AI or similar companies who charge exorbitant amounts to use their APIs. Its time for them they got a reality check
You need to do a deeper dive. Deepseek is not a true AI, it is a distilled AI, which was obviously distilled from ChatGPT, which is why when you asked it (before they fixed the bug) who it was, it said it was ChatGPT. The paper is available, and numerous credible sources are also available on the distillation method they used. While useful for some cases, the reality is without ChatGPT, it does not exist.
Early last week one of it's responses to me was that it was built by OpenAI in the US (it added there is also an office in Montreal). When I switched models, I could never get that response again. During this 'made in USA' phase it had no problem discussing the tiananmen square massacre. Despite all this, I find deepseek to be much more conversational than ChatGPT, it's thought process is more thorough and it shows all of it's thinking upfront, also the results have only very ocassionally been wrong or misguided.
I knew china was already doing really well with ai, it's just the media muffled the research the Chinese have been putting out. Bytedance I believe was the first to make lightning & Hyper models for ultra fast AI image generation speeds to make Live AI painting possible, ArXiv is where I have found a bunch of stuff the Chinese have been posting to and even providing the source code on a lot of stuff.
Thank you deepseek, you made open AI cut prices and give more attempts on free modals. And most importantly thinking. Open ai was giving answers but deepseek taught me how to think solution. Now chatgpt introduced it too. Thats great. For all of it thank you deepseek
This is kind of like the space race. We made it to the moon (supposedly), and then nothing really happened afterward because it was too expensive and took up too much energy.
I love how people talk about "this is worth billions" or "he is worth billions", reminds me of wheelbarrows full of Rubles or Deutsch marks. If we haven't noticed, the dollar is collapsing. I can carry a hundred dollars worth of groceries with one hand .....
Honestly, Moon could've done better with this one, the AI sponsor doesn't help his credibility either. A bit more unbiased research, and he could've painted a better picture that reflects the AI race. I know people in the west won't accept it but the release of Deepseek is arguably the best thing that's happened to the industry, an open source AI produced by reinforcement learning that's on the same level or even better than OpenAI's chatgpt O1 model. Ironicaly Silicon Valley will benefit immensely from this coz it's open-sourced, so they can literally see how the R1 model "thinks". It's true, the cost of the R1 model is up for debate, but still even the real cost is 10X or 20X the 6 million dollar cost, it's still way more cheaper than all the American AIs. It's crazy to think that China is giving the world a real open AI, so developers across the world can alter the model for their own use, and it's on the par with the best Chatgpt model. Most importantly, the entire world can us it for free. Taking into account, the geopolitical political tension between the 2 superpowers, data privacy concern is a real issue. Another beauty of the Deepseek R1 model is that you can run it locally, which may keep all your data on your system. I'm glad Deepseek took the giant leap forward to wake up Big Tech. Google, Microsoft, OpenAI, Apple, all these companies' main focus is gatekeeping the AI industry, racing against each other for a monopoly grip on this game changing technology, crushing all start-ups that could challenge them.
"Greatest lie ever"? not really. This channel is just stating the obvious. Of course the announcement of DeepSeek R1 was meant to disrupt the market led by US tech companies promoting AI. Those were overvalued anyway and the DeepSeek announcement provided the market correction that we need. This channel also failed to mention that DeepSeek is open source and their methodologies are published unlike OpenAI, despite its name, is not open at all. If you are running DeepSeek through their app or cloud service, yes it is censored by China and can be a security risk. However, you can run their model locally even without internet connection or without connecting to their service so in this case security is not an issue. Also because it is open source, anyone can look at it on how the technology works and be able to replicate it. This is a good thing since the technology is no longer monopolized by big tech companies.
I'm running Deepseek locally and it is censored. Ask it about the Tianman Square protests and massacre. You'll see. By running it locally you simply avoid giving your data to China.
Thats is what they are trying to ruin... Cant ruin What is Free to the public... Destroying the Monopoly of Ai training is what they are pissed at.. but they forgot that Human with limitation exceed more than a bottomless pit of resources.
Heres the thing Its not even Good, it makes more mistakes than GPT 3.5, Fake TH-cam videos showing o1 takes over a min to respond ? I use o1 daily it takes almost no time before answering (1 second or less) Not over a min
@@kamelkadri2843not true it has better reasoning than o1 and correctly answered everything I asked that o1 got wrong.
@@kamelkadri2843it's free.
Moon: Deepseek is bad. Anyway this AI that sponsored me is beter.
Yep. Moon can have some very clownish takes at times.
Literally sponsored propaganda. Moon became the thing he hated the most.
Double Standard
@@daudimasinde6280 yea he tends to be anti establishment but leans western establishment when china or Middle East comes in
@ You speak the true true brother.
This video had very little information about how DeepSeek is a lie. This video barely addresses possible lies about the cost of training and the issue of whether R1 is a distilled model using data from OpenAI. This video is an okay overview of the last week of news around DeepSeek, but the topic in the title is barely addressed.
It's open source you can easily ask any person who's proficient in machine learning if it has any resemblance to openai. Also it's arrogant to think that china hasn't been working on any sort of ai model for the past decade.
Because it's not even a "lie" per se. The video is rather useless though, I agree.
DeepSeek got released an entire month BEFORE the crash. The entire premise of this video is invalid. As per usual for Moon.
This same company once lied about making their own computer chips. Turns out they were buying them
Indeed, turning to watchmojo grabage
The problem with ChatGPT is that 99% of it's compute goes into not saying the N word
lmao
This is a much more profound comment than it sounds at first
😭
Are you serious??!! 🤦🏻♀️🤦🏻♀️🤦🏻♀️🤦🏻♀️
@@somejimmydude2121 seeing how Zuckerberg and musk have both given up on screening posts, the safety rails of AI will also probably be withdrawn soon enough.
This whole AI race is very tiring. When is GTA 6 coming out?
its woke
Gta 6 will never get released
Ai is Woke? We r FUCT!
Ask ChatGPT.
Is gonna be woke
Honestly, Moon could've done better with this one, the AI sponsor doesn't help his credibility either. A bit more unbiased research, and he could've painted a better picture that reflects the AI race. I know people in the west won't accept it but the release of Deepseek is arguably the best thing that's happened to the industry, an open source AI produced by reinforcement learning that's on the same level or even better than OpenAI's chatgpt O1 model. Ironicaly Silicon Valley will benefit immensely from this coz it's open-sourced, so they can literally see process of how the R1 model "thinks", and perhaps apply the lessons in their own models.
It's true, the cost of the R1 model is up for debate, but still even if the real cost is 10X or 20X the 6 million dollar cost, it's still way more cheaper than all the American AIs.
It's crazy to think that China is giving the world a real open AI, so developers across the world can alter the model for their own use, and it's on the par with the best Chatgpt model. Most importantly, the entire world can it for free, in contrast to OpenAI's O1 costing hundreds of dollars.
Taking into account, the geopolitical political tension between the 2 superpowers, data privacy concern is a real issue. Another beauty of the Deepseek R1 model is that you can run it locally, which may keep all your data on your system.
I'm glad Deepseek took the giant leap forward to wake up Big Tech. Google, Microsoft, OpenAI, Apple, all these companies' main focus is gatekeeping the AI industry, racing against each other for a monopoly grip on this game changing technology, crushing all start-ups that could challenge them.
Dude, this channel can't even get basic general history right (cf. the video on how USA ruined itself or something like that). Seriously, some basic stuff about The Enlightenment that here in Europe is tought in high school and this channel can't even get that right. Truth is that this is one of those channels that appears smart only to stupid people.
damn lemme just try out that Deepseek and see if its better than chatGPT i didnt know there was a better AI you can just download right now ty for the info lol
I couldn't care less about Sam and his crying over his ChatGPT data being stolen. He stole them from everyone else and monetizing it, he reaped what he sow tbh.
Open-source is good, it always support open competition
So true. Moon is misleading his audience here.
damn lemme just try out that Deepseek and see if its better than chatGPT i didnt know there was a better AI you can just download right now ty for the info lol
I went to the Chinese restaurant and asked what they are cooking today: They said they just cooked OpenAI.
It's not China's fault that the stock market dropped after the release of DeepSeek. The real issue is that many US corporations have invested excessively in artificial intelligence without achieving significant results. As a result, the stocks of several US companies are overvalued, lacking genuine improvements to justify their prices. Additionally, Western regulators have imposed strict controls on the AI industry for several years, hindering its development, as has occurred with the social media sector.
True.
But sure, HighFlyer definitely made a lot from the predictable panic caused by DeepSeek.
my guy has been spying on US corporations
Exactly!
Shut up
DeepSeek got released an entire month BEFORE the crash. This was market manipulation, helped by a few well timed hit pieces.
The entire premise of this video is invalid. As per usual.
If privacy is your concern, simply run it locally.
Exactly. Moon got it all wrong here. I don't know if it's deliberate or not, but it's so biased. He didn't even try to give us the real picture.
Try to run the entirety of chatgpt locally?
@@lxp You can run Deepseek R1 locally
@@lxpyes u can, altho to run the original version you would need a large amount of vram, u can still get the smaller models with more manageable ram requirements altho they won't be as powerful as the original
So as an average person can anyone tell me why I should be concerned about China having my data? It is more worrying for my own government, companies in my own country have my data; they can use it against me. Unsure what the Chinese government will do with any of my data. Sorry if I come across a little naïve.
Bigger lies are Tesla & Nvidia stock prices 😂😅
😊
Journalist : how did they got all these cheaps ?
Elon Must i sold it to them ... obviously they have the cheap i even cut down the price in half
Journalist : ...
Kai-Fu Lee has been working on AI development in China for over 10 years. He worked for Apple and Google respectively before starting his own AI consulting business. The emergence of DeepSeek shouldn’t have been a surprise to anyone, if you know what China has been doing in the last 20 years.
Alexander Wang - the guy in the interview is a bad mount guy. what he said is his elaboration. he even dont need to show any investigation backing for his point.
AI roots for the last 20 years as you said under many names.
You mean stealing from people and companies that actually did the work as they are devoided of any capabilities of innovation and "outside the box" thinking no matter how hard they try? Yeah, effed China. They are solely good at one thing; stealing.
Cool name, Kai-Fu Lee.
You mean he spied at Apple and Google just like the majority of Chinese tech does to America and what they don't get through spying they outright steal a version of it and send it back.
True
Deepseek is actually pretty good.
Keep up the competition.
What is it. Whats the big deal. I never see or hear anyone talking about this nerdy tech sht.
@@ShaquilleOatmeal730 it is as good as OpenAI's best models for a fraction of the price, that's the deal
@@ShaquilleOatmeal730 Its a chat bot you can run on your windows in a LM studio program and is good as chat GPT
It's nothing big, just technology that's revolutionizing the way we live our life. So nothing big at all. @ShaquilleOatmeal730
The fact that chatgpt now allow search on the the free plan and even a feature similar to deepthink😂
DeepSeek - The Greatest Lie Ever Told So where is the Lie ?????
Moon fumbled this one. So biased.
I feel like it will me more often like that, i saw this already on his video about tik tok. He trys to put mor and more his own agenda in his videos i feel like. Sad..
10:30
@@Le_caylemoneyeah I'm starting to noticed he's like one of those lite right wing type channels... not super maga but leaning toward that side and pushing it while trying to not look like it
@@loyisomasango5149 He didnt fumble.
Hes an OP that glows.
Its literally what hes employed and paid to do
And another thing, deepseek said they used less hardware than OpenAI, not believed, a nerd in the US said it believes they have more GPU's without evidence, instantly believed. Americans just can't stand the fact that they aren't at the helm of everything tech.
Commie China got banned from the latest and greatest chips, but use double or triple the amount of the older chips and you can get somewhere...
So why exactly is deepseek "The greatest lie ever told"?
Came for the clickbait, stayed for a really informative video.
Normally; stealing from other blockchains, censorship, and the conspiracy theories owned "Internet of Bodies".
The greatest lie Told was that we need billions of Dollars to create powerfull AI. When a hedge fund created the same thing with less than 1% of that Budget.
This is the problem with society now
My personal belief is that this is not about AI and advancing it, this is akin to a Cold War between the US and China.
We're in a second cold war, but no one wants to say it
It’s only a Cold War because the us keeps turning competition into a cold war
I would say it's more akin to the US being the dominant empire in the world and doing what it can to stop another empire from growing across the planet. Hmm, maybe that is akin to a Cold War. This is very new to human history there is no record of two great empires existing at the same time like this, where one can so greatly influence the other with something as simple as an application on your phone. No Rome vs Greece. or the Persia vs England. One slip-up could change the face of the world. and Unfortunately for the United States our leadership has never been more questionable while China's technology has never been better. I'm getting excited.
If you saw the Chinese military you would laugh yourself to death. They are so amateur compared to the us. They do have great nerds though. Slave nerds but smart nerds nonetheless.
@@pristinegarbagesir2997 which is why they want to fight a tech and economic war.
It was about shorting the tech stocks. Period.
Somebody got it.
Black Friday on tech stocks lmao
Exactly.
It's about being too lazy and/or stupid to think beyond such a simplistic explanation. Go trade some GME and AMC, while more adept people do the thinking.
Dam, the commies made 1T
Moon did you hide the comment criticizing you on AI? I cant find it but i see it
Hahaha same
This is a propaganda channel, it's owned by a larger company I think based in Israel and/or France - there's hundreds like this that pump out content pretending to be individual/authentic youtubers.
See NEWEST
@ Got any proof? Id love to believe this.
@@ForLorNVuLgaR 🙋♂️🐑
The Greatest Lie Ever Told seems too clickbaity
Hardly the greatest lie mate
I can think of at least a dozen worse crimes
Still it's interesting
Perhaps cut out the click bait BS
Religion is easily the greatest lie ever told. Nothing else comes close.
@numerum_bestia entomology the study of language
Shows the word religion literally means
To hold back and to bind
We don't need middlemen to have a connection to source
And religion is the middleman
@@WillAH956Words
At the end of the day more competition like this means better products and cheaper prices for customers. I’m all for it
That’s what is at the surface level
Thats what you'd think an see at surface level
space race 2.0
nope cause the Chinese know how to take create a monopoly and then raise prices only reason temu is cheap.
Americans don't like to lose
When they do, "cheater" is their excuse.
When have we ever lost tho
@jjtehdankplane China is beating you fair and square in almost all domains.
@@jjtehdankplane this time. This tool is extremely useful as a modder.
And it's free
Unlike open ai where I have to pay.
@@jjtehdankplane We have a comedian here.
Not saying it's the case here, but China does have a spectacular track record of IP theft. One should be wary of anything that comes out of an authoritarian country, especially one whose long term results often don't live up to expectations. That said, tech stocks here were a giant circle jerk of constant rising without the evidence to justify it, and a correction was inevitable. Competition is a wonderful thing, and now everyone has a new baseline to work from moving forward.
Lol. Americans. Man, Deepseek is OPEN SOURCE. Your can run it locally in your own machine, host it yourself. No need for their servers. That's the beauty of it. The Deepseek 32B model performs much better than GPT4.
Sorry but not sorry, you're wrong.
It's funny to see them crying
@@lightweight3908The only benchmarks where DeepSeek V3 is surpassed by GPT-4o is SimpleQA and FRAMES. V3 surpasses 4o in all the 20 other benchmarks.
Uh, genus, CENSORSHIP??? Hell, ask Deepseek about the Philippine Sea and they marked as China's claim. Or how about giving the negatives to Mao's regime?
@@finncapistrano9879 What are you talking about?
I'm running the Deepseek model locally and there is no censorship whatsoever. I have ran the two prompts you mentioned were "heavily censored" and have gotten basically the same answer for both. I'm absolutely willing to give you both prompts + outputs from ChatGPT's 4o Model & Deepseek R1's 30B Model so you can compare the output via pastebin or something :) But i'm pretty sure you have no interest in that as you only want to propagate that the release of an open source model comparable to the 4o Model is a bad thing.
by making it open source, deepseek gave everyone the option to host it themselves, or even other companies to host it and offer it to users. that should completely alleviate the privacy/security concerns. unless you use the webUI or the app of deepseek directly. then your data goes to china. be clever about it!
regarding geopolitics: again the big message is that its open source and the methods of training are revealed. that levels the playing field quite significantly in terms of ai world domination, because it gives everyone a new start line to work off of. and in the current state of affairs, this benefits everyone on earth except the US.
regarding nvidia: the training cost of deepseek was way lower than a comparable US model. but that doesnt mean you need less GPUs, it just means you get farther with the same. the stock dip of nvidia was an unfounded investor panic attack. microsoft stock dip was justified, as everyone believed "open"AI was further ahead than they actually are.
First reasonable take ive seen here. If they made deepseek closed and online only, I wouldnt buy it for a bit. but I ran the 32b model and its straight up better than gpt4. seeing how insanely cheap it is to have the full model hosted in america like 2$/1m tokens there is literally no reason to pay openai for sh@t now. if it was all nonsense western AI would have ignored it, like the many other mid models out there.
but tbf the nvidia drop was from two angles, the AI stuff, and the 5000 series being an unmitigated disaster, worse performance than 4 series in some cases, worst generational "uplift" ever?, highest relative price increase (ever??), and worst availability.... EVER????? its actually insane how bad its going for them.
@@mrrooter601 there's a disinformation campagin going on performed by US operatives, OpenAI and china haters making this all about China vs US- but this is actually Open source vs Closed source- and they are banking on common folk not reading the deepseek paper or having technical knowledge to understand the difference
@mrrooter601 good addition, i forgot to mention that inference is also way cheaper with ds.
agree with the 5000series being crap, but gaming is an afterthought for nvidia and investors, so no real influence on the stock price imo
@ nah nvidia is literally one of the most "normie" stocks ever, combine the normies seeing the 5 series being awful, and seeing a selloff and its compounding.
hell the stock started tanking just after the card announcement. now we just started getting the real numbers yesterday for performance too, but its impossible to tell now. Nvidia may not be a gaming company anymore but that doesnt stop idiots for investing in it like it is.
How is bro getting sponsored by AI💀💀😭😭
Bro talks about how AI is a bad thing on stream but clomplainx that ots NOW. Good thing
@@5urufiv bro in coming 2 -3 years, he´s job as continent creator well taken by Ai Software sooooooooooooooo
Deepseek didn't come out of nowhere. The v2 and v3 models were also great and comparable to Open Ai's 4o models. The reason why deepseek is this popular has a lot to do with the reasoning model being so much more efficient and distilled models of R1 that can be run very well on consumer desktops. It doesn't require the Nvidia CUDA architecture. There's a lot more to do with it, but the code efficiency increase is downright phenomenal.
🎯
It is open source, so you can build your own custom agants.
The techniques used are smart, canny in the face of US sanctions.
it has already been superceded, so not to worry !
its almost as if moon is like deep propaganda for the real deep state.
superceded? expand
Ali Baba released Qwen 2.5 , beating DeepSeek’s benchmarks. There are probably lots of up-and-coming models that have been working hard in the background.
@@smkh2890 isnt that more text/creative/artsy, less stem? not a moe right? more 4o than o1? beating deepseekv3 not r1?
@@smkh2890 is Qwen2.5 a competitor to Deepseek-V3 or to the Deepseek-R1 reasoning model???
Also I love how efficient China is at delivering cheap, yet also high quality work. Most things we own are manufactured by China either way, so even if governments try to control the hold they have over our economy, it'll just be a big fat waste of time
"China" and "high-quality" in the same sentence lol...
Yet another dislike for a clickbait Moon video
He's so biased.
So basically the whole videos comes down to, they got to be lying about the price. But then again I don't understand how you can say their lying about the computing power when you can download a local version of deepseek. Totally isolated from Chinese servers. So maybe it was made with the hypothetical 50,000 gpus but clearly they are not need to maintain it.
once the model is train, you dont need to keep training it, and you needed at least a billion to train is the point of this video, so no, deep seek to get to the level we have, didnt cost only less than 10 millions... thats the lie.
@charlestrudel8308 The video does not outright said state 1 billion dollars only stating same guy guesses that it would take 50k gpus to make deepseek then Moon again estimates that 50k of those gpus would cost over 1 billion dollars alone not counting the cost of man hours. Obviously they probably spent over 6 mil. But this video is all just guesses n estimations. Perhaps what's more scary than lying about the price is the idea that the actually found a more efficient way to make AIs.
@@xlegacyzero8878 even if the guess is valid, its still like 25% of the price of the other AI trainning going on, its still quite cheap. and also, a part of the billion i mention is the overhead company expense that are ignored in the initial 6 millions.
@@charlestrudel8308 do you think when we talk about the almost $1 billion dollars to train GPT-o1 we're including the cost of GPUs or salaries??? The cost is measured one-for-one for JUST TRAINING. If we do the same analysis then o1 cost probably tens of billions if not even more!! Please think harder
More people gotta read The Art of War
😆
Space Race: Let's see which superpower nation can create enough force to break past gravity and reach beyond the atmosphere
AI Race: Let's see which superpower nation can generate more words
That is not at all what Ai advancement is about 😭
@@Joe-f1k2c Correct, it's about generating huge, animated catgirl boobies. At the moment USA has the dominant market on animated catgirl boobies but China wants some of the action
🙋♂️🐑🤣
Ai race; let’s build bots that help people, until we build a bot that replaces people.
nah who's have bigger compute capacity
Its open source - thats the whole thing, you can instead of making a fake title like this, copy the code and DO WHATEVER YOU WANT with it....thats like deepseeks all...."greatest lie ever told" haha
Moon lost his plot on this one. Totally wrong. Biased reporting.
This video is poor man , you're becoming a Propaganda now
I like how people believed China's claim at face value. This is the same country where they had a "natural" waterfall at a park, until someone went over the waterfall, only to see a tube spewing water, and post it on the internet.
spot on tbh
or when they said covid couldn't ever come out from the virology lab that is in the city where the breakouts started and it came from bats, out of all animals that have a lot more contact with humans. Bats...
I like how people automatically believe the west and dismiss everything from China.
If you read DeepSeek’s paper, you would know DeepSeek wasn’t built with only 6 million dollars.
what a stupid thing to care about
People also believe any American claim at face value despite the US running the largest Ponzi scheme in history called “social security”. People will constantly believe anything at face value despite a history of lies. Just look at influencer culture.
Comments just saved me 13 minutes and 30 seconds of my life
Played it for background noise but sounded like his usual BS, asks a few questions why, barely or vaguely answers the questions and only just covers the basics with a bunch of filler information. This dudes videos are all the same, seen one seen them all. I get way more context reading the comments
Americans: How dare you steal what I've rightfully stolen?
This video is not well done. Plus it is factually false that this was the greatest lie ever told. And we all know why.
"Greatest lie ever told" was the first title... lmao why did you change the title and thumbnail?
I doubt China did it so cheaply, but I also doubt that U.S. tech firms are going to do anything other than claim they need the most investment dollars possible. At least some of what our tech industry is doing is price gouging for personal gain.
Seriously, why mention that China's economy is collapsing? At the end of your video, I don't see how its population have to do with Deepseek.
The real scandal is the Nvidia sold all the GPUs DeepSeek needed despite the regulations. Probably won't ever be investigated.
The real scandal is the us banning chips because it can’t compete otherwise
That's the "real scandal?" You can't be serious can you?
@@clintholmes2061 He's low information and low critical thought like 80% of the rest of the viewers of these channels who haven't read a book or a few articles of research to save their life.
Never heard of the Black Market eh?
@@coleford6197 Plenty of safe havens around the world to store illicit good no questions asked too.
I wouldent be surprised if it was revealed that that deepseek was made off of smuggled Nvidia chips and more than 7 million dollars.
Most certainly is, not to mention based off of stolen code as well. China doesn’t invent or innovate, they steal and degrade.
they had GPUs for crypto mining and they avoided CUDA so they wouldn't be bound to nvidia cards
Lying about the price doesn't matter. cause the mathematics of it still show that somehow their still about to Chatgpts 4 levels without the $200 price tag and they are somehow still in the green. Possibly proving that American companies have overlooked something vital. it's like if the US invented Helicopers while China invented cars for the average citizens, Why would any person buy a helicoper when the car is 20x cheaper.
Lol Americans
US feel the NEED to ban the export of "high-end" chips to China to slow down it's technological advancements, but China did it anyways. I'm loving China
Or you are just smoking cope
Apart from the points you made, Deepseek is the perfect tool for social engineering and info collection, compared to profit oriented open ai, deepseek doesn't need to generate profit. China will shoulder it anyway.
bro, its open source. you can run the distilled models on a literal smartphone locally.the 32b model runs on 32 gb of ram and a modern gpu, and its on par with gpt4. china cant collect data if you run it locally. china cant collect data if you host it on american servers (if you want the full model), its literally free to download.
obviously if you use their app they are going to collect data on your prompts and use them for further training, thats literally how it works for every AI.
bruh for god's sake its open source that you can run on your own and literally all "free" AIs on the market exist to collect, sell and train on data
average third grader and their mom will sure as hell use "free" Deepseek over paid ChatGPT 4
How about american tools?
@ good news, they explained how they did it. now american startups that want to can follow the open source paper and make their own.
I mean, if you want to, go run the smaller versions of "deepseek" which are actually llama (large language model meta ai aka. facebook) trained on deepseeks outputs.
unless you would only be happy with something that had no chinese involved at all. but lets be real thats pretty uhh ist.
Its not even biased towards the CCPs agenda nearly as much as openais gpt pushes DEI. sure you ask about Tiananmen Square on the app and it will get denied, cause that goes to china directly. and they filter messages going to and from the app.
but asking the (full size) model self hosted, with 100% ownership here in america, and it will give it to you straight. good datasets are hard to find, and since they were actually trying to make the best AI they could, they mostly avoided poisoning it with shitty CCP propaganda, hence the external filtering.
Advancements in AI are going to be a global thing now. since they graciously destroyed "open"AIs grift with the power of open source. Obviously the chinese government has done many things wrong. even if they were even significantly involved here, this they did nothing wrong other than making a better product, for less, and causing a government funded monopoly to seethe.
They PROVED state of the art AI can be 100x more efficient, and that is likely one of the best things the chinese have ever done for the planet. so perhaps we should be grateful they did so and shared how to do it too ya know? maybe give them some credit instead of being a bit ist...
I'm curious how much more effective or less effective it is to western AI in social engineering Chinese propaganda is so transparent in general. Although western propaganda has gotten pretty bad itself. I think the more aware of social engineering the general public becomes the less effective it is. Although if one comes up with a new idea for pushing it the learning curve will ensure many fall for it before word can get around. Although how many new methods of pushing the same establishment goals can they come up with? Imo there's only so many ways to convince people to blindly support the establishment.
AI taking AI's jobs is wild
US is losing and will keep losing.
While US were focusing on wars and oil and Middle East and terrorism that most Americans actually think was a inside job, China was focused on technology and finance and global expansion. I think China has shown the world it is right now perhaps the most stable and reasonable country in the world and I say this in a context where the US and Europe are only focused on wars and China had the perfect opportunity to go ape like everyone else right now and invade Taiwan. But they didn't. They have been working for peace while channeling their energy to solve issues like these - a decaying superpower that is crumbling and wants to prevent them to get chips. I wonder if persons realise how insanely dumb that is. In the world we live in there will always be someone ready to smuggle those marvelous chips everyone is talking about right now the same way back in the 80's there was a nut job of a Russian pilot willing to steal a MIG mid training and fly all the way to Hokkaido risking his life because the fuel barely made it and when he got there he didn't know where and how to land. But the US got their hands on that MIG didn't they?
good thing about deepseek, its open source so anyone can make there own version and dont need to worry about data being stolen.
This whole video is starting wrong already. DEEPSEEK was release two days BEFORE trump announced StarGate
Moon downfall 2025
It's bad. So much bias on this video, makes me want to question all his other content I've consumed.
The greatest lie in history? Really. think you’ll find that title goes to the existence of God!
Moon seems to be headed in the same direction as Jake Tran
Indeed. I haven't watched Jake Tran in years. Moon is losing the plot.
As soon as I saw the deepseek release, I immediately assumed all of this even though I knew none of the specifics. I’ve been following US/China relations for years but recently through the perspectives and narratives from natives and people who lived there and know it without the government censorship. The biggest problem with AI is the mindset controlling it because once a superior model emerges, it’s up to the humans who are utilizing it to decide what its morality will hinge on. Nobody wants China to be in that position.
Stop twisting the truth. I've got DeepSeek and it runs flawlessly on my old GTX, the model is open-source so you can test it yourself. You do realise that, ultimately, it was the banks/hedge-funds etc. that cashed in the biggest by shorting tech stocks. They just wait for industry-disruptions to make big moves, that's the game.
Okay Ziganwu...
@ who tf is that
@@zd3f027xh5w Bruh, a foreigner who simps Chinese trends and Marxism
@@finncapistrano9879 ok
China crashes the market for Nvidia chips so that China can buy them at a cheaper prices. Chinese is playing 4D chess right there, allegedly.
Guys remember when the American rich spoke of greatness of the free market (and now block the export of nvidia chips with sweaty brow)... how capitalism breeds innovation (and see all the planned obsolence and dlc:fication of purchased products) and other funny memes... These guys speak so much bs that it's probably hard to pull occasionally a decent breath in.
American capitalists fired US workers and moved their factories to China and Mexico!
It's so ironic.
The reason it used so much less resources was just stealing from openai. But once theres no one to steal from the resources needed will become the same. The good thing is they released an open source version, which should become an infistry standard
is it about the lie we tell ourselves in an effort to cope?
Totally. Love tbe meltdown of fake tech gurus in US like Elon who have never even been to the DeepSeek repo on github and read their whitepapers.
Crazy how good deepseek is but then cant answer what happened on Tiananmen Square in 1989
DeepSeek more like Deepstate
WHAT DO YOU MEAN ARE YOU MAKING A JOKE???
lol meanwhile OpenAI hired literal deep state actors
@@PuppetMasterdaath144 Good question.
The only thing that matters is: Does it work? If it does, everything else is uninteresting!
Deep Seek = Stupid.
Deep Thought = GOAT.
Deep throat = BEST
Deep Dish = Chef's Kiss
Deep Shit = U.S.A.
Deep Throat = Classic
After I watching the whole video, I'm left with the conclusion that everything is a big MAYBE. a propaganda video at best, and I'm disappointed.
This channel is good at peddling propaganda but usually it's more subtle. Still don't know if it's deliberate or just some dumb kid who exposed himself to a specific kind of content too much. Hope the latter, but the more I watch from this channel (usually avoid it like the plague) the more I'm scared it's the former.
The more competition on the Ai market, the better. $200 a month is ludicrous. No one company should have a monopoly. Just look at the good that did Nvidia. They made a killing and a half.
The name of the game is optimazation more so than power. If things keep going the way they are we will be able to use an ultra powerful ai on a dinky phone!
People in the comments acting like they knew all along 😅
And then you look through their history to find that isn't the case. In almost every single comment lmao.
The fact that black rock lost something, nothing short of amazing. The fact the 800s are just 100s programmed to run not as good, nvidea could face a huge investigation. No smoke without fire. Oooof
Moon just changed...
The propaganda and bias videos are too much... -1 subscriber
Ok, dude. We get it. CCP got their accounts to run here to post the same thing 1000 different ways to intimidate the channel into not saying anything.
@@WillmobilePlus It's not even about CCP bots attacking content creators. We're saying this out of love for the channel. I've been watching "Moon" for years, but I've never been so disgusted by the bias in this video. I'm fortunate enough to being familiar to the AI race to AGI, and Deepseek in particular.
Moon has always been one of those channels that are a breath of fresh air, and great analysis of the status quo. But this is most biased video I've watched in this channel, without any intention of giving the audience a clear reflection of the truth. I'm so disappointed, and unsure if this was deliberate or not but Moon dropped the ball on this one.
@@loyisomasango5149 Yeah, sure.
Spare us the playact. He was critical of some thing out of China, and like on other videos of the same, here you guys come with just variations of the same BS post about how "he dropped the ball" and "I'm so disappointed blah blah blah". The usual crap Chinese bot farm posters type.
7:45 - Did you get the correct picture of Liang Wenfeng? I think it might be another person.
Capitalist propaganda goes brrrrrrr
13:10
It’s dumb to question if this app spy’s on you , just like it’s dumb to think if facebook spies on you
AI will make GTA 6 before Rockstar will
Uhhh this is not a good argument as we can all see the open source model and it can quite literally be trained on simple h20s for 1/1000th the cost. Like you dont even need h100s or h200s!!
another factor is that deepseek is able to run so lean because it focuses on the highest quality data for training
Data, as with almost all fields, 80% of the value comes from 20% of the set. Deepseek only uses the most valuable 20%, and can therefore run light and train fast. But how does it know what data is valuable? Easy, it steals and cherry picks it.
Short term you look brilliant, long term you will hit a wall pretty quick
It has better architecture too.
@@curious_one1156 no point talking to people who didn't read the paper and can only regurgitate other misinformed opinion they only heard about
I believe US tech bois gonna take a lot of advantage of deepseek being open source. Don't forget to thank China 😂
This video is a classic case of someone trying to milk a topic by presenting an alternative viewpoint just to create drama and get some of the attention.
To sum this up: *'Yo guys. This new thing that is trending rn because it's functional, free, was cheap to create, is open source etc... is... bad.. actually.. somehow... Idk how, but lemme just pretend like it is....*
*Btw, let me introduce the sponsor of another AI company for this video...'*
0:23 It should be $500000 Million (= $500 billion) instead of $5000 Million (which is $5 billion). SMH American Math.
this is just like a research paper. It wouldnt cost much effort if there are already studies supporting yours. Especially when its the same exact topic.
get help "moon"
“Ne-Vidia” 😂
He should know better
If he can’t pronounce nvidia.. moon might actually be a wiki shill.
AI generated voice
@ i could believe that
I got so triggered by that 😂
Its not sputnik moment, capable open source AI were available way before of the deepseek hype. This black swan event was artificially created.
What happened to this channel?
Terrible
The moment I saw their spent budget I immediately thought: "Oh, they have IT & Tech Sweatshops in there now."💀
Another bigoted rant video.
This is the first video of Moon I'm giving thumbs down. Focus shouldnt be on the cost of deepseek, but on Open AI or similar companies who charge exorbitant amounts to use their APIs. Its time for them they got a reality check
Exactly. He got it all wrong. I don't know if the misleading is intentional or what, but this is disappointing from him
so u tell me, this was their side project to shorting, but they just beat entire west most advance LLM? LMAO
You need to do a deeper dive. Deepseek is not a true AI, it is a distilled AI, which was obviously distilled from ChatGPT, which is why when you asked it (before they fixed the bug) who it was, it said it was ChatGPT. The paper is available, and numerous credible sources are also available on the distillation method they used. While useful for some cases, the reality is without ChatGPT, it does not exist.
so what when one man invented a machine and others used that knowledge to give a better efficient machine than the previous one with cheap cost
And without the data scraped for free from the Internet, there would be no chatgpt. So ... your point is what exactly?.
Early last week one of it's responses to me was that it was built by OpenAI in the US (it added there is also an office in Montreal). When I switched models, I could never get that response again. During this 'made in USA' phase it had no problem discussing the tiananmen square massacre. Despite all this, I find deepseek to be much more conversational than ChatGPT, it's thought process is more thorough and it shows all of it's thinking upfront, also the results have only very ocassionally been wrong or misguided.
@@bobbysani2126 Which could have never happened without building the first one.......clearly not a lot of thinkers out there.......
@@randenpatterson297 i stole from the internet and i will accuse other company stealing from me
Not lot of thinkers out there
The fact that you said Nuh-Vidia offends me 😂
I knew china was already doing really well with ai, it's just the media muffled the research the Chinese have been putting out.
Bytedance I believe was the first to make lightning & Hyper models for ultra fast AI image generation speeds to make Live AI painting possible, ArXiv is where I have found a bunch of stuff the Chinese have been posting to and even providing the source code on a lot of stuff.
lol china copied and stole that's all they did
Thank you deepseek, you made open AI cut prices and give more attempts on free modals. And most importantly thinking. Open ai was giving answers but deepseek taught me how to think solution. Now chatgpt introduced it too. Thats great. For all of it thank you deepseek
This is kind of like the space race. We made it to the moon (supposedly), and then nothing really happened afterward because it was too expensive and took up too much energy.
Another but of nebulous waffle... I remember when this channel was actually insightful.
Thought I was the only one who was seeing this. It's so bad. Half of his points are either untrue or misleading🤮
U going down 🌑❤🤣 from ☀️☝️🇧🇻💕
DeepSeek is the best. Oh and it is free.
Great video but your voice is sometimes difficult to hear and understand.
That's because it's an AI voiceover.
@@marzupalami Not sure if you are being serious or not
I love how people talk about "this is worth billions" or "he is worth billions", reminds me of wheelbarrows full of Rubles or Deutsch marks. If we haven't noticed, the dollar is collapsing. I can carry a hundred dollars worth of groceries with one hand .....
DEEPSEEK …❌
DEEP CCP✅✅
Open AI ❌️❌️❌️
Closed CIA, US Armed Forces, Freedom spreader (Oil seeker) ✅️✅️✅️
US dikRider indian ✅
@@lly_09 US dikRider indian ❌❌
UR SISTER(MY CHINESE WIFE) IS INDIAN DIKRider✅✅✅
Honestly, Moon could've done better with this one, the AI sponsor doesn't help his credibility either. A bit more unbiased research, and he could've painted a better picture that reflects the AI race. I know people in the west won't accept it but the release of Deepseek is arguably the best thing that's happened to the industry, an open source AI produced by reinforcement learning that's on the same level or even better than OpenAI's chatgpt O1 model. Ironicaly Silicon Valley will benefit immensely from this coz it's open-sourced, so they can literally see how the R1 model "thinks".
It's true, the cost of the R1 model is up for debate, but still even the real cost is 10X or 20X the 6 million dollar cost, it's still way more cheaper than all the American AIs.
It's crazy to think that China is giving the world a real open AI, so developers across the world can alter the model for their own use, and it's on the par with the best Chatgpt model. Most importantly, the entire world can us it for free.
Taking into account, the geopolitical political tension between the 2 superpowers, data privacy concern is a real issue. Another beauty of the Deepseek R1 model is that you can run it locally, which may keep all your data on your system.
I'm glad Deepseek took the giant leap forward to wake up Big Tech. Google, Microsoft, OpenAI, Apple, all these companies' main focus is gatekeeping the AI industry, racing against each other for a monopoly grip on this game changing technology, crushing all start-ups that could challenge them.
Its always cute to see the copium. Like those videos who says China economic total collapse since 2002
Moon your amazing, Been watching you for years, keep it up!
Criticize AI but Sponsored by AI? The bill can't be that high😂
Lmfao 🙏💀
Same hype they made on Black Myth wukong 😂 now no one talking about BMW it's only use on benchmark videos😂