Is DeepSeek something to worry about?

[viljy]

I find it hard to believe in the "we made a chatbot, that is better than ChatGPT for less than 10 million dollars" theory.

It could be. Considering that they used a different approach to model training (let's call it effective). There is also a very significant factor for the huge expenditure of investors' money - the more you spend (inefficiently), the more you can demand in the future. Or, for example, Ilya Sutskever got a billion for his company, and now he's looking for another 20 billion. At the same time, it was immediately stated that before the creation of AGI, his company would not produce anything. And it's indefinitely long.

[Sanitough]

If you trust them, you provide them with data, so you’re willing to take the risk.

But what’s the difference between using ChatGPT and DeepSeek if the concern is about data collection? Both operate similarly, but they’re developed in different countries.

People use ChatGPT without worrying about data collection, yet they’re concerned about DeepSeek doing the same thing. That just means they trust the U.S. over China when it comes to handling data.

Most of top Tech companies are from U.S like Google, Facebook, Microsoft, Amazon and list goes on. It's very likely true that U.S has access to data hosted on servers of these companies. In current century, data is the new oil. Once countries fight for oil, these days countries have strong concerns over there data reaching in hands of rival countries.
Moving forward, AI is the new emerging technology. What's stunning about DeepSeek is there claim that they have invested only $6 million to train there final model, that is much cheaper then OpenAI’s o1. If DeepSeek take over the AI world then it clearly means that China has upper hand in AI compared to US, something not tolerable by US.

I just got that idea as I've been watching Netflix for ages. In the movie Snowden, it shows how the US government spies on people. Maybe that's why they want to lead the tech industry to use that power for spying. And they might just be trying to make China look bad, even though they've probably been doing it forever.

here's the movie. https://www.imdb.com/title/tt3774114/  - Snowden | Official Trailer [HD] | Open Road Films

[Hispo]

First of all, if you think that China stealing your data is a new thing, you need to wake up and realize that they are taking as much data from you as they can, hell they would take your mail and password if they can, they are all fine for it, you are their opponent, and they are a dictatorship, they want to control not just their own citizens, but everywhere else too, that's what a dictatorship does.

...

I have a personal theory on them being collecting data from people all over the world thanks to the dominance Chinese handheld devices have had in these lastest years, specially in developing markets in Asia, Latin America and Africa. I am talking about brands like Xiaomi, Tecno, Poco, Infinix, etc. They use their own modified version of Android and also their own launcher, so it is the worst of both worlds: while we are being spied on by Google (because we continue to use Google's software) are also being spied on by the Chineae government, thanks to the cheap hardware they have managed to put out in the market of smartphones and other devices.

So far the world had the chance to wake up and put a stop to massive surveillance when Snowden disclosed what the government of the USA and it's allies were doing with millions of unaware people, but we decided not to sacrifice any comfort for the sake of freedom and right to privacy, thus companies in China and USA feel entitled to continue gathering our information for their benefit.

[Lucius]

I'm more concerned about the excessive use of AI, which has become so "normal" in a very short time that it seems to me that in the next ten years or so (maybe even sooner) people will start to take it for granted and will use their brains less and less. At the same time, I am particularly concerned that artificial intelligence for military purposes is increasingly being talked about (and tested) to the extent that drones and even fighter planes will be controlled by AI and will make autonomous decisions.

What we are seeing now is clearly a race to see who will develop a better AI, and the way things look, it seems that what we see in science fiction movies is what awaits us in the future. People are obviously so stupid that they want to create something smarter than themselves - only when that smarter thing realizes that it exists, I don't doubt for a moment that it will turn against its creators.

[moneystery]

america is just exaggerating the impact of deepseek's development. isn't this just a normal business competition? and i think that in any industry, the emergence of competitors is natural and needed to make the industry more advanced, developed, and more affordable for users.

but america is afraid that their monopoly on the global ai industry will be disrupted, and the valuation of their ai companies and chip manufacturers will decrease due to the presence of deepseek and other competitors.

it shouldn't just be deepseek, but companies from other countries should also emerge as competitors to make the ai ​​industry less centralized and make ai technology even smarter and more affordable for all.

[lovesmayfamilis]

Why are AIs created? Now it sounds like new technologies that help people, however, the fact that people become stupid when reading social networks, and two-line posts without the ability to reason independently, since there is always someone at hand who will say better and write better, has already been proven many times. Why is all this being created? Those who can go further and see what inaction and laziness lead to, will understand that AI is a good weapon to control people. Everything has already been decided and proven by machines, you just need to believe what AI writes. Some people are so lazy and unable to read for more than five minutes that many believe what is offered to them.

[STT]

AI represents ease of use so it will propagate very fast I think, the amount of good utility left unused is unfortunate.   If we consider wealthy countries and high wages existing in those economies then AI has a place if it can better make use of peoples time.

The years Ive been waiting for computers to take over repeated actions required on a computer is far too long.  AI is actually late, it should have been a regular thing as much as 10 to 20 years ago.  Actual expertise and genuine usage of tools will still be a thing but endless form filling for people shouldnt be as common as it is, we have so many gaps between systems where people must waste their time navigating.

AI first coming to call centres probably says alot, much time and effort wasted in relaying basic information to customers or allowing them to access routine account data is where AI should thrive best.  It will improve many companies and the experience for customers no end.   WIth deepseek itself making AI less elitist and more available I do not find a fault especially.

Path of least resistance and ease of use rating highly for product use and competitiveness in a market place has to mean AI demand is genuine, massive and I believe long lasting; I'm glad its finally here.

[Lucius]

~snip~
Path of least resistance and ease of use rating highly for product use and competitiveness in a market place has to mean AI demand is genuine, massive and I believe long lasting; I'm glad its finally here.

I would rather say that the AI ​​craze has been imposed as something that everyone needs to participate in, otherwise they will remain stupid and fall behind others. Now it all comes down to whether my AI is better than yours, and how much money did you invest in it and how much did I so that our stocks could skyrocket and we could all get rich.

It would be false to say that there is no benefit from AI, because as far as I can see, it is of great help in healthcare when it comes to diagnostics - but the fact that people today become creative with the help of AI is pure stupidity and deceiving themselves.

[Yucky]

I don't see how DeepSeek is something to worry about. One reason why people are now talking about DeepSeek is because it blew up. There are other AIs people use. There is Claude, there is Gemini. Different AI, but just that DeepSeek happened to make the fame at some point in time. Now, has anything  deadly really happened since all the AI have been in use? There are more AIs in existence, than you know. These tools were programmed to help you work better and they have limitations.

There are certain requests you will make on DeepSeek or AI tools in general, and they will tell you "sorry, they are unable to assist you because they were not programmed with that I formation, and the have information from 2023 max ". So If you really don't want your data or anything there, then do not try them at all. That's just how I see it. Asides from AI inclusion, there are other factors that also disrupt a country's economy like international relations that I feel we should even pay attention to more than the AI sector.

[WatChe]

It could be. Considering that they used a different approach to model training (let's call it effective). There is also a very significant factor for the huge expenditure of investors' money - the more you spend (inefficiently), the more you can demand in the future. Or, for example, Ilya Sutskever got a billion for his company, and now he's looking for another 20 billion. At the same time, it was immediately stated that before the creation of AGI, his company would not produce anything. And it's indefinitely long.

With time if DeepSeek reveals more details about there inner workings like hardware, algorithms etc only then it will be possible to believe that have indeed developed an AI Chatbot that costs 6 million $. If an AI powered Chatbot can be developed in 6 million $ then we will see more such projects coming in future. For now we have no other option but to believe that China indeed has developed a ChatBot in 6 million $ budget and China has gained AI supremacy over US.

Just like there are conspiracy theories about Apollo 11 project that it was fake and created by US in 1969 to beat Russian space superiority. These days big powers fight to gain technology superiority over each other rather then fighting a war.   

[coupable]

It would be false to say that there is no benefit from AI, because as far as I can see, it is of great help in healthcare when it comes to diagnostics - but the fact that people today become creative with the help of AI is pure stupidity and deceiving themselves.

AI can change life in the future in many ways, it can impact industries, jobs, healthcare, education, entertainment, etc. For example, AI can help improve the quality of healthcare services, as smart machines can diagnose diseases and recommend the best treatment plan.

[STT]

people today become creative with the help of AI

Every task is better done by humans, I dont think I would argue computers are creative and people 2nd to that.  Its really about the case of amplifying results and replication of a process by a semi intelligent bot.

We are used to stupid computers and it doesnt have to be that way, even worse users lack knowledge and training of how to use the tool in front of them so its wasted.   If AI improves that it'll be massive.

[The Sceptical Chymist]

What concerned me, though, isn't the economic threat or the supposed drop in the cryptocurrency market, but a video I saw a few days ago on Instagram and TikTok, showing that traffic is being sent to China and on Alibaba Cloud. On the one hand, it sounds plausible because their servers are situated there, right? On the other hand, though, it is possible that it's sending our personal data.

Yep, that seems to be the concern of other countries as well and was the main point of the article you linked to a far as I can tell.  It's a valid concern, too, if DeepSeek is or turns out to be some sort of global data collection mechanism that would give China access to sensitive data and/or allow it to become a threat by way of having an intelligence advantage (and I'm talking about intelligence in the military/political sense).

Then again, my paranoid brain is also worried about my own country spying on its citizens and its own data collection methods, as well as the power of big tech companies doing likewise.  We're in 1984, folks.

People use ChatGPT without worrying about data collection, yet they’re concerned about DeepSeek doing the same thing. That just means they trust the U.S. over China when it comes to handling data.

Correct me if I'm wrong, but the only thing these AI programs do is collect data.  That's the whole point, isn't it?  To integrate what you write with everything else that's been put into its algorithm and come up with an output?  That's how AI programs learn, no?

[shinratensei_]

It could be. Considering that they used a different approach to model training (let's call it effective). There is also a very significant factor for the huge expenditure of investors' money - the more you spend (inefficiently), the more you can demand in the future. Or, for example, Ilya Sutskever got a billion for his company, and now he's looking for another 20 billion. At the same time, it was immediately stated that before the creation of AGI, his company would not produce anything. And it's indefinitely long.

With time if DeepSeek reveals more details about there inner workings like hardware, algorithms etc only then it will be possible to believe that have indeed developed an AI Chatbot that costs 6 million $. If an AI powered Chatbot can be developed in 6 million $ then we will see more such projects coming in future. For now we have no other option but to believe that China indeed has developed a ChatBot in 6 million $ budget and China has gained AI supremacy over US.

Just like there are conspiracy theories about Apollo 11 project that it was fake and created by US in 1969 to beat Russian space superiority. These days big powers fight to gain technology superiority over each other rather then fighting a war.  

I thought deepseek is open source and everything already revealed, afaik, the $6m cost coming from the calculation of computing hour, the hour needed to train the LLM with the array of H800 that they owned, so this estimation doesn't account into the salary of the employee and the fact that deepseek bought the GPU not rent.

as for other thing, pretty much deepseek has published the inner workings of their model, even some university already using distillation method for their own LLM.

[pooya87]

I thought deepseek is open source and everything already revealed,

The other important components of AI technology is the database they build and use. That is private and I don't think any of the AI developers have ever published it for the public to see.

[Lucius]

~snip~
Correct me if I'm wrong, but the only thing these AI programs do is collect data.  That's the whole point, isn't it?  To integrate what you write with everything else that's been put into its algorithm and come up with an output?  That's how AI programs learn, no?

We can say that AI is still somewhat limited because at least it has not been officially released into the wild - but I think there are models that are much more advanced and that already have full access to online information. Some people are convinced that such AI is capable of making decisions much faster and more efficiently than humans can, and this is clearly a war being waged between the US and China. We should not forget that most analysts predict that wars in the future will be fought with the help of autonomous machines powered by super AI.

If you look at what they're already doing in China, it's not hard to imagine - fully autonomous machines are building highways and they seem to be doing it much better than humans.

China has successfully resurfaced a 158-kilometer stretch of the Beijing-Hong Kong-Macao Expressway using drones and autonomous robots. This project highlights the transformative potential of AI and autonomous machinery in the field of road construction.

The project employed drones equipped with high-resolution cameras to survey the highway and identify repair needs. These drones provided precise data, which autonomous robots used to execute the resurfacing with remarkable accuracy. This method not only significantly reduced the time required for completion but also minimized traffic disruptions, ensuring a smoother flow for motorists.

[bitLeap]

If you trust them, you provide them with data, so you’re willing to take the risk.

But what’s the difference between using ChatGPT and DeepSeek if the concern is about data collection? Both operate similarly, but they’re developed in different countries.

People use ChatGPT without worrying about data collection, yet they’re concerned about DeepSeek doing the same thing. That just means they trust the U.S. over China when it comes to handling data.

How dangerous is the data you mean? Has there been an incident that has successfully threatened data by ChatGPT or DeepSeek. Because what I know is that this sophistication is not about threats but more about the fear spread by its competitors. For example ChatGPT spreads fear to DeepSeek users and vice versa, that's all that is happening now. The existence of new competitors that threaten its business sector, challenges from countries that have wanted to shift the position of the global market from the start. And for those of us who do not have a role, we only need to take advantage of the facilities they provide without having to join the cold war.

[Lida93]

What are your thoughts? Is it an economic threat? Do you also believe that it's collecting any data we submit to it?

Indeed the USA economy is strong enough to withstand any threat to it by other economies. However on this note I presume that the DeepSeek AI model does have an equal match with that of the USA for it to be able to significantly distort the stock market down to cryptocurrency market leading to a fall in prices of all coins.

But if we careful analyse this development it doesn't really make for a threat as we or some persons might be envisaging. DeepSeek AI model was just a shock to the market. I think there are users that would still glue to the ChatGPT of the USA despite the discrepancy in cost to that of DeepSeek for personal reasons. DeepSeek model as an alternative for those who can't meet with the cost of the former is no bad idea as it creates room for choice for end users.

In the collection of data I don't think that DeepSeek is anything different from the USA ChatGPT because they both collect and make use of data as made available online.

[Swordsoffreedom]

~

How dangerous is the data you mean? Has there been an incident that has successfully threatened data by ChatGPT or DeepSeek. Because what I know is that this sophistication is not about threats but more about the fear spread by its competitors. For example ChatGPT spreads fear to DeepSeek users and vice versa, that's all that is happening now. The existence of new competitors that threaten its business sector, challenges from countries that have wanted to shift the position of the global market from the start. And for those of us who do not have a role, we only need to take advantage of the facilities they provide without having to join the cold war.

After 22 days of release, Deepseek has more than 22 million users and has become the fastest growing AI application in the world. And so far, I have not heard any reports or allegations that they are collecting user data and using it illegally or any users being threatened for using Deepseek. There are no reports or evidence other than news from their competitors. Likewise, no OpenAI users have ever been in danger or threatened.

You are right, all these fears are just because we have entered a war that is not ours but a competition between suppliers and they are trying to manipulate us to their advantage.

[tygeade]

Most of top Tech companies are from U.S like Google, Facebook, Microsoft, Amazon and list goes on. It's very likely true that U.S has access to data hosted on servers of these companies. In current century, data is the new oil. Once countries fight for oil, these days countries have strong concerns over there data reaching in hands of rival countries.
Moving forward, AI is the new emerging technology. What's stunning about DeepSeek is there claim that they have invested only $6 million to train there final model, that is much cheaper then OpenAI’s o1. If DeepSeek take over the AI world then it clearly means that China has upper hand in AI compared to US, something not tolerable by US.

I just got that idea as I've been watching Netflix for ages. In the movie Snowden, it shows how the US government spies on people. Maybe that's why they want to lead the tech industry to use that power for spying. And they might just be trying to make China look bad, even though they've probably been doing it forever.

here's the movie. https://www.imdb.com/title/tt3774114/  - Snowden | Official Trailer [HD] | Open Road Films

Maybe? I mean that would be the most ideal way to use to spy on people if they can figure out a way to use it for spying, Snowden isn't the only one, we have seen plenty of other stuff where USA and other nations all used to spy on their own people, every single nation does that whenever they can, and USA will use AI for that too if they can. Of course this depends on the agency that will spy on people, and how they can use AI to do it, while I understand that people think they will, we should also realize that they are not that capable, we are talking about agencies that are bloated and usually a failure by this point, they are not as strong as they used to be.

This is why we need to be careful but also do not have to worry about that easily. FBI and CIA isn't what they used to be, they are still of course spying both their own citizens and also other nations citizens if they deem them a threat as well, it's just that they are not as good as they used to be at it.