Welcome to DU! The truly grassroots left-of-center political community where regular people, not algorithms, drive the discussions and set the standards. Join the community: Create a free account Support DU (and get rid of ads!): Become a Star Member Latest Breaking News Editorials & Other Articles General Discussion The DU Lounge All Forums Issue Forums Culture Forums Alliance Forums Region Forums Support Forums Help & Search

highplainsdem

(54,628 posts)
Sun Jan 26, 2025, 01:59 PM Jan 26

"The race for 'AI Supremacy' is over, at least for now, and the U.S. didn't win." It just became a tie with China.

The sentence in quotes in the thread title is from AI expert and critic Gary Marcus's latest Substack, just this morning. Link and excerpts from that below, and links to other articles and a video.

I've been so focused on politics this past week that I gave almost no attention to AI news, or Marcus's tweets and Bluesky posts. I also missed usonian's post about this on Thursday - https://www.democraticunderground.com/10971911 - in the Open Source and Free Software group here. Mea culpa. I should have posted about this in LBN days ago. Will try to explain why this is a BFD now.

Gary Marcus's Substack about this, which I hope you'll read in its entirety:

https://garymarcus.substack.com/p/the-race-for-ai-supremacy-is-over

The race for "AI Supremacy" is over, at least for now, and the U.S. didn't win. Over the last few weeks, two companies in China released three impressive papers that annihilated any pretense that the US was decisively ahead. In late December, a company called DeepSeek, apparently initially built for quantitative trading rather than LLMs, produced a nearly state-of-the-art model that required only roughly 1/50th of the training costs of previous models, — instantly putting them in the big leagues with American companies like OpenAI, Google, and Anthropic, both in terms of performance and innovation. A couple weeks later, they followed up with a competitive (though not fully adequate) alternative to OpenAI's o1, called r1. Because it is more forthcoming in its internal process than o1, many researchers are already preferring it to OpenAI's o1 (which had been introduced to much fanfare in September 2024). And then ByteDance (parent company of TikTok) dropped a third bombshell, a new model that is even cheaper. Yesterday, a Hong Kong lab added yet a fourth advance, making a passable though less powerful version of r1 with even less training data.

-snip-

Realizations about Deepseek are already reverberating. Altman has already cut prices twice in the last few days, and we already know from his own testimony that OpenAI has been losing money on ChatGPT Pro; as Grady Booch observed on X, “Now he’ll be losing money faster.”

Nvidia could soon take a serious hit, too, for two reasons. First, the DeepSeek results suggest that large language models might be trained far more efficiently going forward. Nvidia has basically been getting rich selling (exceptionally well-designed) shovels in the midst of gold rush, but may suddenly face a world in which people suddenly require far fewer shovels.

Second, the DeepSeek results suggest that it is possible to get by without NVidia's top-of-the-line chips. Devices shipped in armored cars and traded on the black market might suddenly seem "nice-to-haves" rather than must-haves.

-snip-


From Investor's Business Daily this morning:

https://www.investors.com/market-trend/stock-market-today/dow-jones-futures-meta-tesla-microsoft-buy-points-earnings-due/

AI infrastructure stocks got a boost last week from the big Stargate project announced by President Trump. But, despite Meta's big capital spending plans, many AI hardware stocks fell Friday, including Nvidia. Why? Chinese AI lab DeepSeek released a powerful AI model that it claims cost just $6 million to build, using less-powerful chips. If true — or partially true — that could undermine demand for super-pricey AI chips and other gear.



Articles:

https://mashable.com/article/deepseek-ai-hands-on

https://venturebeat.com/ai/deepseek-r1s-bold-bet-on-reinforcement-learning-how-it-outpaced-openai-at-3-of-the-cost/

https://www.wired.com/story/deepseek-china-model-ai/

https://www.forbes.com/sites/petercohan/2025/01/26/nvidia-stock-may-fall-as-deepseeks-amazing-ai-model-disrupts-openai/


CNBC video on DeepSeek:

15 replies = new reply since forum marked as read
Highlight: NoneDon't highlight anything 5 newestHighlight 5 most recent replies

madville

(7,571 posts)
2. CNBC headline is an AI stock rout today
Mon Jan 27, 2025, 09:35 AM
Jan 27

Bottom about to fall out I guess, oh well. The rich will just sell now and buy it back cheaper, then run it back up again at some point, and repeat.

blogslug

(38,761 posts)
3. Ed Zitron helps me understand
Mon Jan 27, 2025, 09:46 AM
Jan 27
https://bsky.app/profile/edzitron.com/post/3lgpwdmpmdk2x

in reply to:

Seya @sey4star.bsky.social‬

Can someone explain it to me like I'm stupid?


Ed Zitron @edzitron.com‬

The AI bubble was inflated based on the idea that we need bigger models that both are trained and run on bigger and even larger GPUs. A company came along that has undermined the narrative - ways both substantive and questionable - and now the market panicked that $200bn got wasted on AI capex


highplainsdem

(54,628 posts)
5. Yes. Sam Altman talking about getting $7 TRILLION in funding, as he was last year, is now
Mon Jan 27, 2025, 09:55 AM
Jan 27

in a different investment universe.

Johnny2X2X

(22,507 posts)
6. Boy, this seems premature when AI doesn't really exist yet
Mon Jan 27, 2025, 09:57 AM
Jan 27

We have search engines that can organize data a little better, that's not AI.

Renew Deal

(83,637 posts)
8. Deepseek has shot to the top of App Stores
Mon Jan 27, 2025, 10:01 AM
Jan 27

I don’t know where the downloads are coming from, but I doubt they’re all in China.

Renew Deal

(83,637 posts)
13. It took the #1 spot yesterday
Mon Jan 27, 2025, 11:44 AM
Jan 27

And remains there. Rednote was #1 last week, so we'll see how long this lasts.

Hugin

(35,823 posts)
9. Back in ancient times (2021 - 2022) when I was experimenting with a variety of pre-ChatGPT LLMs.
Mon Jan 27, 2025, 10:43 AM
Jan 27

I noticed some things that were directly related to the size of a training set. Which may still be applicable.

Personally, I was using some terms to differentiate the effects.

The larger sets, which I called “deep” have many bifurcations or “branches”. The vast majority of which, the conditions would rarely arise to be “traversed”. On those rare occasions are when a generative AI response could result in what has been called nonsense or a “hallucination”.

The flaw, as I saw it, with these smaller training sets is that they tend to be what I call, “brittle”. They contain mostly the highest probability branches to be traversed. When a rare condition occurs, instead of hallucinations, they “break”. I never came up with a solution to these breaks other than to feed a reset to the “trunk”. These resets were scripted prompts. I suppose it’s possible that there’s other methods for implementing a partial reset. But, I didn’t spend any time on it.

Once again, all of these observations about generative AI and it’s fitness are subjective and by no means objective. Does a smaller training set create a better AI? Who says it’s better and why. I never cease to be amused by the answers to those simple questions.

Yavin4

(37,110 posts)
12. While America is trying to re-live its Jim Crow past, China owns the future.
Mon Jan 27, 2025, 11:13 AM
Jan 27

We have lost the future.

kysrsoze

(6,261 posts)
14. This is a very good thing for our planet, as these technologies would require a tiny fraction of energy use projections.
Mon Jan 27, 2025, 01:08 PM
Jan 27

I've been worried about the power draw for both this and the crypto-currency global pyramid scheme. President Caligula even suggested building coal-plants to serve the extra power "need."

tulipsandroses

(6,914 posts)
15. trump said he would give 500 billion to the tech bros to develop AI
Mon Jan 27, 2025, 01:09 PM
Jan 27

So will DOGE speak out against this? Eloon has beef with Altman, maybe he will?

Latest Discussions»General Discussion»"The race for 'AI Suprema...