It is really not that hard to understand why Chinese AI like Deep Seek is a lot cheaper, and more effective, at least, in the Chinese language domain. Chinese researchers and engineers focus on model and logic backed up by vast amounts of data collected by many big companies like Tencent. Alibaba, Xiaomi, etc, while the U.S. counterparts rely heavily on chips capacity and capabilities for computing. There is far more human intelligence involved in Deep Seek than ChatGTP. Human intelligence is lot cheaper and efficient and effective than hardware. The Western media pretend not to understand this simple fact because they don’t know how to compete with China and they don’t know how to deal with the consequences in the human intelligence domain. Cooperation with China is the only viable option for the U.S., so that the U.S. would know a lot more about what China is doing, and vice versa, for a better world.
I was with you until you made it sound as if china is in a superior position and the US should just accept this. Don't bet against freedom of speech, it allows for creativity not available to others ;)
If DeepSeek helps build an open source global platform for collective terrestrial intelligence, CTI, I'm all for it. The platform needs to be able to merge, deduplicate, fact check, and aggregate the knowledge and sentiment expressed in public conversations with billions of people around the world. People could access the platform with their phones.
Are these guys engineers? If engineers, then don't talk BS. China probably has tens thousands engineers, hundreds companies, if not more, actively working on AI as we speak.
I just can't help but cringe when they talk about how deepseek distilled openai models with no evidence. That would be the most inefficient way to generate data for a 671 billion parameter model, especially considering the breakthroughs attained by deepseek are around efficiency.
@@rayauxey The evidence is when you ask R1 what model it is, it sometimes replies "GPT-4". And I think their paper also say they distilled a larger model.
@@andrasbiro3007 In the paper, they distilled Deepseek R1 which is a big model of 671 b parameters to smaller open models like llama and qwen. They distilled thier own model.
@ But they got the training data from somewhere fro the big model. And OpenAI claims it was generated through their API. Also, the Deepseek models someimes claim they are ChatGPT. That's not necessarily proof, but strong evidence.
The key is risk analysis. The markets are and has always been unpredictable. Thus the important point is not how much one might make but what one could lose. trading and holding on speculative stocks/Crypto and even great stocks can test ones ability in the arena of our ability to control ones loss aversion bias…buy high sell low…The market is overvalued by almost all measures like the CAPE and Buffett indicator. Speculation is high with things like meme stocks (are they dead yet?), our kitty dude, Ai dreaming and all.... It is all about how much one can lose during times like these. Does that mean don't invest, of course not but one must access the risk and have a way out if the tide goes out.....I've been engaged in active trading and managed to grow a nest egg of around 2.6B'tc to a decent 24B'tc....I'm especially grateful to Seren Wintersun, whose deep expertise and traditional trading acumen have been invaluable in this challenging, ever-evolving financial landscape.
This is exactly what Big time investors are talking about, not just you. key thing here is to always trades with a good strategy even if the price goes up or down.
Seren goes deeper than just looking at surface-level trends.. she explores technical, fundamental, and sentiment analysis, offering a comprehensive perspective on the market.
Nick, I agree that network effects are critical. However, price matter as well. Hence monitization of these LLMs will be different. High monthly subscriptions won't work anymore with a commoditized product since price sensitive consumers would move away from a more expensive incumbent tool toward a cheaper newer player.
Brad was correct to the extent that the extra GPUs will eventually used but DeepSeek shows that there is more you can do with efficiency improvements and you didn't need to spend $50,000 per chip like they have been paying so far. The GPU price could come down by 10X in the next few years and would be much cheaper to to build the capacity then. .
Excellent points, everyone is missing this. It’s the equivalent of the race for hp in auto engines. US = bigger engine more horsepower and German is more horsepower through engineering and efficiency on same engine.
Great points as Brett puts it... why would any of the current AI major players slow down their development due to DeepSeek's results with less resources? In addition, DeepSeek is also using Nvidia chips... IMHO, this knee jerk negative reaction was an over-reaction...
Yeah, that’s the big question. On one hand, it has some impressive applications, but on the other, there’s always the worry about misuse-whether it’s privacy issues or even financial manipulation.
I think the financial aspect is what worries me the most. If this technology can predict or manipulate markets in ways we don’t fully understand, that could create serious instability.
That’s why having a solid financial plan is more important than ever. With all these uncertainties, working with someone like James Brendan McCall can help ensure your investments and retirement funds are well-protected.
I think you need to crystallize first what DeepSeek is doing differently in algorithm, what speed up factor that gives. It’s not about whether the brute force approach of throwing more GPU will cease, it’s about how well it fares against the alleged algorithmic speed up, if any.
@Bret There's no AI vendor lock-in if the competing AI is smart enough to learned everything your current AI can do and replicate it. You just adopt the new AI, tell it to replace all the tasks your current AI does and your done.
Hollywood loves producing movies about some quiet Caucasian men or women who invented something big enough to change the world. Now, it's a real-life story when more non Caucasians around the world did the impossible, too ! 80% of the global citizens are watching more miracles outside Hollywood's movies 😂
Jai Hind. All the Oligarch closed AI models will be turn into DeepShits. DeepSeek is something like the light weight Wing Chun Kungfu grand-mistress Luo FuLi versus all the Baddies ultra-heavy Hyperscaling Oligarchs in AI world.
DeepSeek's AI breakthrough bypasses industry-standard CUDA for some functions, uses Nvidia's assembly-like PTX programming instead. DeepSeek reconfigured Nvidia's H800 GPUs: out of 132 streaming multiprocessors, it allocated 20 for server-to-server communication, possibly for compressing and decompressing data to overcome connectivity limitations of the processor and speed up transactions. To maximize performance, DeepSeek also implemented advanced pipeline algorithms, possibly by making extra fine thread/warp-level adjustments.
This is a stupid question and stupid brain fart, DEEPSEEK is the best Ai model there is : free or not and its open source, go take a deep breath and enjoy this wonderful innovation instead piss around with stupid content
It is really not that hard to understand why Chinese AI like Deep Seek is a lot cheaper, and more effective, at least, in the Chinese language domain. Chinese researchers and engineers focus on model and logic backed up by vast amounts of data collected by many big companies like Tencent. Alibaba, Xiaomi, etc, while the U.S. counterparts rely heavily on chips capacity and capabilities for computing. There is far more human intelligence involved in Deep Seek than ChatGTP. Human intelligence is lot cheaper and efficient and effective than hardware. The Western media pretend not to understand this simple fact because they don’t know how to compete with China and they don’t know how to deal with the consequences in the human intelligence domain. Cooperation with China is the only viable option for the U.S., so that the U.S. would know a lot more about what China is doing, and vice versa, for a better world.
I was with you until you made it sound as if china is in a superior position and the US should just accept this. Don't bet against freedom of speech, it allows for creativity not available to others ;)
It is NOT fair to assume Chinese gov is involved. Plus u have no idea how LLM is trained. For god's sake.
These people are brainwashed. DeepSeek vs shallow puddles.😊
If DeepSeek helps build an open source global platform for collective terrestrial intelligence, CTI, I'm all for it. The platform needs to be able to merge, deduplicate, fact check, and aggregate the knowledge and sentiment expressed in public conversations with billions of people around the world. People could access the platform with their phones.
Of all the conversations that's available on UA-cam regarding this topic. This conversation is actually focused on the right points. 👍👏👏👏
Are these guys engineers? If engineers, then don't talk BS. China probably has tens thousands engineers, hundreds companies, if not more, actively working on AI as we speak.
I just can't help but cringe when they talk about how deepseek distilled openai models with no evidence. That would be the most inefficient way to generate data for a 671 billion parameter model, especially considering the breakthroughs attained by deepseek are around efficiency.
@@rayauxey
The evidence is when you ask R1 what model it is, it sometimes replies "GPT-4".
And I think their paper also say they distilled a larger model.
the fact these people supporting deepseek are th same people against the ticktoc ban is just funny
@@andrasbiro3007 In the paper, they distilled Deepseek R1 which is a big model of 671 b parameters to smaller open models like llama and qwen. They distilled thier own model.
@
But they got the training data from somewhere fro the big model. And OpenAI claims it was generated through their API.
Also, the Deepseek models someimes claim they are ChatGPT. That's not necessarily proof, but strong evidence.
The key is risk analysis. The markets are and has always been unpredictable. Thus the important point is not how much one might make but what one could lose. trading and holding on speculative stocks/Crypto and even great stocks can test ones ability in the arena of our ability to control ones loss aversion bias…buy high sell low…The market is overvalued by almost all measures like the CAPE and Buffett indicator. Speculation is high with things like meme stocks (are they dead yet?), our kitty dude, Ai dreaming and all.... It is all about how much one can lose during times like these. Does that mean don't invest, of course not but one must access the risk and have a way out if the tide goes out.....I've been engaged in active trading and managed to grow a nest egg of around 2.6B'tc to a decent 24B'tc....I'm especially grateful to Seren Wintersun, whose deep expertise and traditional trading acumen have been invaluable in this challenging, ever-evolving financial landscape.
SHE IS ON TELE GRAM.
@Serenwintersun
This is exactly what Big time investors are talking about, not just you. key thing here is to always trades with a good strategy even if the price goes up or down.
Seren goes deeper than just looking at surface-level trends.. she explores technical, fundamental, and sentiment analysis, offering a comprehensive perspective on the market.
Great skills and knowledge about the market. I enjoy full profits and easy withdrawal with no complains,
Best show EVER! Thanks guys.
Nick, I agree that network effects are critical. However, price matter as well. Hence monitization of these LLMs will be different. High monthly subscriptions won't work anymore with a commoditized product since price sensitive consumers would move away from a more expensive incumbent tool toward a cheaper newer player.
Ark, there are 10 more China Chatgpt in line to go online and in English. You guys have been living in yr own bubble.
Brad was correct to the extent that the extra GPUs will eventually used but DeepSeek shows that there is more you can do with efficiency improvements and you didn't need to spend $50,000 per chip like they have been paying so far. The GPU price could come down by 10X in the next few years and would be much cheaper to to build the capacity then. .
Excellent points, everyone is missing this. It’s the equivalent of the race for hp in auto engines. US = bigger engine more horsepower and German is more horsepower through engineering and efficiency on same engine.
Great points as Brett puts it... why would any of the current AI major players slow down their development due to DeepSeek's results with less resources? In addition, DeepSeek is also using Nvidia chips... IMHO, this knee jerk negative reaction was an over-reaction...
I’ve been hearing a lot about Deep Seek lately. It’s fascinating technology, but should we be concerned about its potential risks?
Yeah, that’s the big question. On one hand, it has some impressive applications, but on the other, there’s always the worry about misuse-whether it’s privacy issues or even financial manipulation.
I think the financial aspect is what worries me the most. If this technology can predict or manipulate markets in ways we don’t fully understand, that could create serious instability.
That’s why having a solid financial plan is more important than ever. With all these uncertainties, working with someone like James Brendan McCall can help ensure your investments and retirement funds are well-protected.
Google his full name
JAMES BRENDAN MCCALL
I think you need to crystallize first what DeepSeek is doing differently in algorithm, what speed up factor that gives. It’s not about whether the brute force approach of throwing more GPU will cease, it’s about how well it fares against the alleged algorithmic speed up, if any.
broke tech bros' monoply
Competition is good for consumers...
China vs China in AI 2026..
😂😂😂😂😂
And your track record is absolute shit
@Bret There's no AI vendor lock-in if the competing AI is smart enough to learned everything your current AI can do and replicate it. You just adopt the new AI, tell it to replace all the tasks your current AI does and your done.
Hollywood loves producing movies about some quiet Caucasian men or women who invented something big enough to change the world. Now, it's a real-life story when more non Caucasians around the world did the impossible, too ! 80% of the global citizens are watching more miracles outside Hollywood's movies 😂
Breatt id the Legend ❤
My Mac OS is running Linux under the hood?
Dah
After Solar, EV & now it’s time for AI that’s Chinese innovation pattern believe it or not
Proberly a stupid question. Does open AI o1 work in Chinese.
Not stupid, but yes. o1 can converse in Chinese and many other languages.
Jai Hind. All the Oligarch closed AI models will be turn into DeepShits. DeepSeek is something like the light weight Wing Chun Kungfu grand-mistress Luo FuLi versus all the Baddies ultra-heavy Hyperscaling Oligarchs in AI world.
Yes
No we should not worry.......
Hot Take: Google has most upside and downside from AI of the Mag 7.
So you are saying DeepSeek did a Vulcan mind-meld on llama and the rest who did all the work vetting their data. Saves money.
Breakin News: World leaders are congratulating CHINA DEEPSEEK RI for helping the world for FREE! ❤❤❤
Expext more realb open-source firms world wide
Off topic, congrats for ARKK hitting a 52 week high today, the 30th. We are on our way to a new ATH!🎉
This is putting me to sleep 🥱
sour grapes bruh
NO
Please add subtitles
Deepseek suck 😂
DeepSeek's AI breakthrough bypasses industry-standard CUDA for some functions, uses Nvidia's assembly-like PTX programming instead. DeepSeek reconfigured Nvidia's H800 GPUs: out of 132 streaming multiprocessors, it allocated 20 for server-to-server communication, possibly for compressing and decompressing data to overcome connectivity limitations of the processor and speed up transactions. To maximize performance, DeepSeek also implemented advanced pipeline algorithms, possibly by making extra fine thread/warp-level adjustments.
This is a stupid question and stupid brain fart, DEEPSEEK is the best Ai model there is : free or not and its open source, go take a deep breath and enjoy this wonderful innovation instead piss around with stupid content
What's your problem?
Agree with you, this sounds like ppl who are caught short