toucana Posted Monday at 03:18 PM Posted Monday at 03:18 PM A new Chinese AI app called ‘DeepSeek R1’ (深度求索 - shēndù qiúsuǒ ) has roiled the AI stockmarket sector to the tune of $1 Trillion, and has taken the #1 position in Apple’s App Store, ahead of ChatGPT and other competing AI products. https://www.theguardian.com/business/2025/jan/27/tech-shares-asia-europe-fall-china-ai-deepseek What has attracted so much attention from analysts and investors is that the Hangzhou-based startup who created DeepSeek claim to have spent little more than $6 Million in developing the product, and they did so without the help of Nvidia’s most advanced H100 chips which have been banned by USA from export to China since September 2022. DeepSeek was created using the less powerful H800 chips developed by Nvidia for the Chinese market (which were subsequently also banned from export to China last October). Quote DeepSeek’s success at building an advanced AI model without access to the most cutting-edge US technology has raised concerns about the efficacy of Washington’s attempts to stymie China’s hi-tech sector. Marc Andreessen, a leading US venture capitalist, compared the launch of DeepSeek’s R1 model to a pivotal moment in the US-USSR space race, posting on X on Sunday that it was AI’s “Sputnik moment”. According to DeepSeek, its R1 model outperforms OpenAI’s o1-mini model across “various benchmarks”, while research by Artificial Analysis puts it above models developed by Google, Meta and Anthropic in terms of overall quality.
Genady Posted Monday at 03:29 PM Posted Monday at 03:29 PM This is the AI that failed my tests, e.g., https://www.scienceforums.net/topic/135352-help-to-share-beauty-of-math-with-more-people/#findComment-1283477 https://www.scienceforums.net/topic/135377-are-tangent-points-allowed-in-ruler-and-compass-constructions/#findComment-1283623
exchemist Posted Monday at 03:46 PM Posted Monday at 03:46 PM 22 minutes ago, toucana said: A new Chinese AI app called ‘DeepSeek R1’ (深度求索 - shēndù qiúsuǒ ) has roiled the AI stockmarket sector to the tune of $1 Trillion, and has taken the #1 position in Apple’s App Store, ahead of ChatGPT and other competing AI products. https://www.theguardian.com/business/2025/jan/27/tech-shares-asia-europe-fall-china-ai-deepseek What has attracted so much attention from analysts and investors is that the Hangzhou-based startup who created DeepSeek claim to have spent little more than $6 Million in developing the product, and they did so without the help of Nvidia’s most advanced H100 chips which have been banned by USA from export to China since September 2022. DeepSeek was created using the less powerful H800 chips developed by Nvidia for the Chinese market (which were subsequently also banned from export to China last October). Oh marvellous, the end of the US monopoly on chatbots to fill the internet with shit. Now we can have Chinese shit too. Splendid, splendid. 1
CharonY Posted Monday at 04:20 PM Posted Monday at 04:20 PM 54 minutes ago, swansont said: AI is the latest stock bubble. That is pretty much it. Stock values are to a large degree dependent on investor's sentiments and expectations. I.e. value is a derivative of factors rather than a hard-wired to products. To me, it seems that stock bubbles is not that different to things going viral on the internet. And I suspect the internet has added volatility to some degree, especially in the tech sector. 34 minutes ago, exchemist said: Oh marvellous, the end of the US monopoly on chatbots to fill the internet with shit. Now we can have Chinese shit too. Splendid, splendid. I am pretty sure that Chinese actors would be happy to use American shit to further their agenda, too. After all, they use twitter a lot outside of China.
exchemist Posted Monday at 04:36 PM Posted Monday at 04:36 PM 12 minutes ago, CharonY said: That is pretty much it. Stock values are to a large degree dependent on investor's sentiments and expectations. I.e. value is a derivative of factors rather than a hard-wired to products. To me, it seems that stock bubbles is not that different to things going viral on the internet. And I suspect the internet has added volatility to some degree, especially in the tech sector. I am pretty sure that Chinese actors would be happy to use American shit to further their agenda, too. After all, they use twitter a lot outside of China. Yeah, actually I didn't mean the deliberate, Bannon-style "Flood the zone with shit", so much as just the junk and bad information that is piling up. I read just the other day, in the Financial Times, about social media being full of AI-generated "slop". It's not just us people on science forums who can see how bad these chatbots are.
CharonY Posted Monday at 04:44 PM Posted Monday at 04:44 PM I think both are related to various degree. Also, considering that quite a few folks TikTok, facebook, twitter, etc. as their main source of information, bot-generated content has been on the menu for quite a while. While they may have originally been human-generated, the short format, plus the propensity of certain folks to keep sharing that nonsense, made them highly prolific regardless of their source. Of course, the amount will continue to increase with AI but I think one of the bigger issues is that they will look more sophisticated (especially images).
swansont Posted Monday at 05:23 PM Posted Monday at 05:23 PM 1 hour ago, toucana said: What has attracted so much attention from analysts and investors is that the Hangzhou-based startup who created DeepSeek claim to have spent little more than $6 Million in developing the product, and they did so without the help of Nvidia’s most advanced H100 chips which have been banned by USA from export to China since September 2022. DeepSeek was created using the less powerful H800 chips developed by Nvidia for the Chinese market (which were subsequently also banned from export to China last October). Doesn’t look like they inquired about whether or not the AI was trained by stealing copyrighted material
CharonY Posted Monday at 07:14 PM Posted Monday at 07:14 PM 1 hour ago, swansont said: Doesn’t look like they inquired about whether or not the AI was trained by stealing copyrighted material To be honest, I suspect that most AI companies don't want folks to look that closely. I would actually be more surprised if they didn't as most are treating it a bit like Wild West (as evidenced by the numerous ongoing lawsuits).
toucana Posted Tuesday at 12:14 AM Author Posted Tuesday at 12:14 AM I found an amusing video where a guy asked DeepSeek R1 to solve Lagarias’s Elementary Version of the Riemann Hypothesis - one the seven unsolved Millenium Prize Problems - which the Clay Mathematics Institute offered a prize of one million dollars for a solution to, back in the year 2000. https://www.youtube.com/watch?v=KQsQaORVhfA
iNow Posted Tuesday at 02:56 AM Posted Tuesday at 02:56 AM 11 hours ago, exchemist said: Now we can have Chinese shit too. Part of the reason it crashed the market is bc it outperformed the best state of the art model from openAI at 3% the cost and they also open sourced it where performance was confirmed on a local installation (I.e. could not be gaming the responses). It also showed how the massive investments in GPUs may be overblown since they did all this on weaker machines (the only ones they could get bc of sanctions). Necessity is truly the mother of invention 10 hours ago, CharonY said: one of the bigger issues is that they will look more sophisticated (especially images) Elon’s xAI model Grok is one of the best performing. It sure is a good thing he has zero nefarious intentions /sarcasm
CharonY Posted Tuesday at 03:07 AM Posted Tuesday at 03:07 AM Oh yey. The US tried to export democracy with guns. Now they try fascism with bots. What a time to have a brain.
exchemist Posted Tuesday at 09:30 AM Posted Tuesday at 09:30 AM 6 hours ago, CharonY said: Oh yey. The US tried to export democracy with guns. Now they try fascism with bots. What a time to have a brain. Good point. Whatever happened to Bill Kristol and the rest of the Project for a New American Century crowd? I expect they now find themselves, to their amazement, on what Trumpies regard as the Libtard Left! For instance I see that old walrus John Bolton has just had his security protection (against being assassinated by Iran) removed, purely because he has criticised Trump.
swansont Posted Tuesday at 06:38 PM Posted Tuesday at 06:38 PM 9 hours ago, exchemist said: Good point. Whatever happened to Bill Kristol and the rest of the Project for a New American Century crowd? I expect they now find themselves, to their amazement, on what Trumpies regard as the Libtard Left! For instance I see that old walrus John Bolton has just had his security protection (against being assassinated by Iran) removed, purely because he has criticised Trump. From a MAGA perspective, probably, since they seem to have a “if you’re not for us you’re against us” attitude. But not from traditional policy perspective from those on the left. And I don’t think it’s to their amazement. They see Trump pretty much for who he is.
exchemist Posted Tuesday at 07:10 PM Posted Tuesday at 07:10 PM 29 minutes ago, swansont said: From a MAGA perspective, probably, since they seem to have a “if you’re not for us you’re against us” attitude. But not from traditional policy perspective from those on the left. And I don’t think it’s to their amazement. They see Trump pretty much for who he is. I've just seen this clip, of a press conference given by Senator Chris Murphy this afternoon, which strikes me as spot-on. Trump is acting like a Stuart era monarch : 1
Recommended Posts