Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

On DeepSeek and export controls

darioamodei.com

Post reply on HN
Print view View original

141–150 of 185 posts

Re: On DeepSeek and export controls

#141

Post by breakitmakeit »

hellojesus wrote: post #125
breakitmakeit wrote: post #119

Earlier quoted context omitted.

And thus is shown that legal reasoning is not equivalent to moral reasoning

In my opinion, it would be more moral to allow charity to cover healthcare for low wealth individuals over a government provided solution. But even supposing a government funded solution is best, why push to the feds what could be solved by the state?

That is... bonkers. At the risk of feeding a troll, I find it difficult to take the argument that voluntary contributions to public health through charity is more "moral", which in the case of health care I would argue is equivalent to "effective", than a system implemented by an organization with enormous power (the government) which at least theoretically has a direct duty to its stake holders (voters), who in turn have the power to enact change in that system if it's not serving them? (by voting)

They just kind of have to give up their power and hope?

Buddy, one cursory glance at history will show that hoping gets you nowhere.

Re: On DeepSeek and export controls

#142

Post by xtracto »

jddj wrote: post #3
epoch_100 wrote: post #2

> DeepSeek does not "do for $6M what cost US AI companies billions". I can only speak for Anthropic, but Claude 3.5 Sonnet is a mid-sized model that cost a few $10M's to train (I won't give an exact number). Also, 3.5 Sonnet was not trained in any way that involved a larger or more expensive model (contrary to some rumors). Wow!

The narrative is running away in every direction. There are popular infographics floating around today which explain to the layman that deepseek invented MoE.

Shit... I was just looking at the DeepSeekMoE paper, almost exactly from 1 year ago:

https://arxiv.org/html/2401.06066v1

The red star in the first chart is so funny.

Re: On DeepSeek and export controls

#143

Post by MiiMe19 »

suraci wrote: post #136
MiiMe19 wrote: post #98

Earlier quoted context omitted.

[flagged]

if you're chinese you'll know some chinese were sterilized because they _didn't_ believe in a religion muslims were pardoned of child policy because of their ethnics, it's a privilege of them people can not imagine something they never saw, try dig your own history, you'll find why it's always about religonal genocide, religonal sterilization, and religonal concentration camps

I am not talking about the one child policy here. I am talking about the ongoing sterilizations of Uyghurs.

Re: On DeepSeek and export controls

#144

Post by dang »

BizarreByte wrote: post #23
throwaway_32u10 wrote: post #20

I'm kind of tired of the entire narrative that "US good, everyone else bad", as if only US deserves to hold powerful technology because it will be used for "the greater good", rather than employ it in military applications, as if Anthropic didn't partner with Palantir. And it amazes me how many people can't seem to see past the "US is good, everyone else are bad" smoke mirror.

[flagged]

Please don't take HN threads further into flamewar hell, and certainly not nationalistic flamewar hell. It's repetitive and evokes worse from others.

https://news.ycombinator.com/newsguidelines.html

Re: On DeepSeek and export controls

#145

Post by inglor_cz »

suraci wrote: post #114
inglor_cz wrote: post #31

Earlier quoted context omitted.

Multipolar world is probably going to have multiple wars on the "edges of control", similar to the one now raging in Ukraine. It may also have some positives, too, but situations with several similarly sized powers competing for primacy have historically been pretty bloody. Unlike Apple and Google, the US, China, Russia etc. have military forces at their disposal.

well, from the historical perspective, it's more easy(or equally hard) to achieve a peaceful multipolar world than a eternal unipolar world i mean, all 'rule-based international order' died in history, the Holy Roman Empire, the First Empire of France, and the ancient Chinese empire died in every 300 years even there is no China, even the empire doesn't implode, will africans and south americans be willing to mining…

Can you name a single period of peaceful multipolar world in situations where the Great Powers could actually reach one another militarily?

I mean, the fact that the Romans and the Chinese never clashed over future Kazakhstan is not a proof of a peaceful multipolar world. It is simply a corollary of the fact that neither empire was able to expand into the steppe.

Re: On DeepSeek and export controls

#146

Post by kristjansson »

rcarmo wrote: post #112
kristjansson wrote: post #105

Earlier quoted context omitted.

To be sure, DeepSeek did great work, and this is a bit aside from TFA. But the PTX thing is a bit of meme? What do we think torch.compile and triton and llvm's nvptx backend are doing under the hood? The warp-specialization thing quoted in [1] cites to a _2014_ paper[2] out of Stanford ... [2]: https://dl.acm.org/doi/10.1145/2555243.2555258

Yeah, well, it's not _just_ PTX. Think about what you would do if you had to work in a resource-constrained system (that's a mindset I closely relate to since I still do C++ for MCUs, and it makes you dig _under_ the libraries to save resources).

Totally, they did great work under their constraints. Training in FP8, the MLA thing they introduce in DeepSeek-V2, etc. I just take particular issue with the attention the PTX thing is getting because (a) it's not like other labs don't do stuff like that and (b) it doesn't contribute nearly as much to their outcome as the other algorithmic and operational improvements they've made.

Re: On DeepSeek and export controls

#147

Post by inglor_cz »

scilro wrote: post #134
inglor_cz wrote: post #31

Earlier quoted context omitted.

Multipolar world is probably going to have multiple wars on the "edges of control", similar to the one now raging in Ukraine. It may also have some positives, too, but situations with several similarly sized powers competing for primacy have historically been pretty bloody. Unlike Apple and Google, the US, China, Russia etc. have military forces at their disposal.

US unipolar hegemony has averted an all-out war, but it has also been very bloody, or at least immiserating, for people at its periphery. Multipolarity doesn't imply we go back to the 1910s. The idea would be to strengthen multilateral institutions that put a check on things like the World Wars.

Humans are pretty much the same as they were in the 1910s, or the entire history before.

If there is anything that keeps us from tearing each other's throats out, it is a) democracy, which in most countries makes going into a war of aggression somewhat harder (I know, the US is a huge exception, but Trump I. was partially elected on the basis of Clinton being perceived as a war hawk), and b) the intuition that weapons are now so destructive that there is nothing to win, except scorched earth.

Still that didn't stop Putin from launching his war; miscalculation such as his ("the enemy is a paper tiger and will fold immediately") is very possible even today.

IIRC the same delusion led Saddam to invade Iran in 1980.

Re: On DeepSeek and export controls

#148

Post by suraci »

gadtfly wrote: post #103
gadtfly wrote: post #84

Earlier quoted context omitted.

^ This is publicly new information, and the 2nd part especially contradicts consequential rumours that had been all-but-cemented in closely-following outsiders' understanding of Sonnet and Anthropic. Completely aside from anything else in this article.

Also, though it's not "new information": "Making AI that is smarter than almost all humans at almost all things [...] is most likely to happen in 2026-2027." continues to sail over everybody's head, not a single comment about it, even to shit on it. People will continue to act as though they are literally blind to this, as though they literally don't see it.

we're getting used to it

and, personally, i think, if any CEO in this industry dare to say "we won't get a super AI in 2028"

many people will be disappointed, some people will be scared, and one prisident be pissed off

Re: On DeepSeek and export controls

#149

Post by CuriouslyC »

MiiMe19 wrote: post #126
CuriouslyC wrote: post #120

Earlier quoted context omitted.

Funny you mention fascists. As for China being a concentration camp gang, that's not going to matter when their former big brother/ally is just relentlessly bullying them and trying to extract all sorts of concessions under duress, and the alternative has demonstrated to be a reliable ally and partner.

Once again, I don't really care if Europe decides that is who they want to be friends with. If your country thinks that trade is more important than the country currently directly committing a modern holocaust, I don't want us to be allied with you at all.

Ironically, if Trump keeps running this playbook, it's going to end with a USA/Russia/Israel bloc and a Europe/LatAm/Africa/China bloc. Which one of those sounds like the evil empire to you?

Re: On DeepSeek and export controls

#150

Post by itishappy »

daft_pink wrote: post #15

I just want to ask if when we talk about export controls on AI chips, is this going to create export controls on general consumer goods in the near future. Given Moore’s law and the efficiency that will clearly come from optimizing chips for AI and competition increasing the amount of VRAM on these devices to run models locally. Is creating an export regime today going to mean that in 3 or 4 years general smart phone…

I don't expect so. Export controls are aimed at chips needed for training, not inference.

The H20 with 96GB of memory is currently available in China. We're a ways off from restricting consumer devices.

Post reply on HN