What is the state of Chinese AI research?
post by Ratios · 2022-05-31T10:05:48.138Z · LW · GW · 2 commentsThis is a question post.
Contents
Answers 8 Kayden 6 Perhaps None 2 comments
It's not a secret that China invests heavily in AI research, yet almost all the new developments discussed publicly originated in the US.
What is the reason for this? Can anyone maybe share interesting insights into what happens in China?
Answers
ChinAI takes a different approach: it bets on the proposition that for many of these issues, the people with the most knowledge and insight are Chinese people themselves who are sharing their insights in Chinese. Through translating articles and documents from government departments, think tanks, traditional media, and newer forms of “self-media,” etc., ChinAI provides a unique look into the intersection between a country that is changing the world and a technology that is doing the same.
ChinAI might be of interest to you.
China, overrated probably - I'm worried about signs that Chinese research is going stealth in an arms race. On the other hand, all of the samples from things like CogView2 or Pangu or Wudao have generally been underwhelming, and further, Xi seems to be doing his level best to wreck the Chinese high-tech economy and funnel research into shortsighted national-security considerations like better Uighur oppression, so even though they've started concealing exascale-class systems, it may not matter. This will be especially true if Xi really is insane enough to invade Taiwan.
Gwern has some insights in this post. Probably more stuff to be found on his website or twitter feed.
↑ comment by David Johnston (david-johnston) · 2022-06-02T03:07:07.989Z · LW(p) · GW(p)
Are the unranked Chinese exascale systems relevant for AI research, or is it more that if they've built 2-3 such systems semi-stealthily, they might also be building AI-focused compute capacity too?
↑ comment by burmesetheater (burmesetheaterwide) · 2022-05-31T19:42:52.182Z · LW(p) · GW(p)
This part of the analysis is both biased and extremely superficial. It may also be correct, but one might give low credence at face value.
Replies from: burmesetheaterwide, gwern, david-johnston↑ comment by burmesetheater (burmesetheaterwide) · 2022-06-01T19:41:53.918Z · LW(p) · GW(p)
A few reflections on a tragically wrong comment:
- What does what I think matter? Make argument, don't invoke myself if not necessary.
- It seemed obvious why the analysis is biased, but maybe this isn't the case, and maybe more info should have been provided. Mostly the concern here is over wording like "Xi seems to be doing his level best to wreck the Chinese high-tech economy" and "shortsighted national-security considerations" and "Uighur oppression" and (to paraphrase) maybe their leader is insane enough to invade Taiwan. To have all of these pop up in a single paragraph that's supposed to be about AI raises some red flags. Does it need to be explained why?
- Calling the analysis superficial without explicitly justifying this is problematic, particularly as the response is even more superficial.
Yesterday I saw that gwern's response was heavily upvoted but didn't understand why; maybe it is part of a mechanism to keep people below a certain intellectual threshold off the site.
↑ comment by gwern · 2022-05-31T20:32:49.512Z · LW(p) · GW(p)
This part of the thread is both biased and extremely superficial. It may also be correct, but I would give low credence at face value.
Replies from: M. Y. Zuo↑ comment by M. Y. Zuo · 2022-06-01T04:26:07.963Z · LW(p) · GW(p)
This is remarkably childish for you Gwern, idk why nobody else in the last 8 hours has called you out on it. Even if burmesetheater was an outright troll, which does not appear to be the case, your still using lame mimicry that would barely be acceptable in the worst parts of reddit.
It’s actually so out of character for you that I’m now having second doubts about your claims, along with quite a few other folks in all likelihood, given your long posting record of mostly coherent writing.
Replies from: gwern, lc↑ comment by gwern · 2022-06-01T14:08:07.573Z · LW(p) · GW(p)
My point, proven by construction, was that burmesetheater's comment was completely information-free and generic and extremely superficial and applicable to anything whatsoever, pure argument from authority; and yet it was sitting at >4 with no criticism. I could have simply downvoted it, but then burmesetheater (and all his upvoters) might go around thinking that his brave truthtelling was being censored and that LW is in even more dire need of his insights than he thought. So I gave it a politer and better response than it deserved.
Strange to call it 'childish' or 'disrespectful' when I put much more thought into it than he put into his original comment, nor is it out of character for me; I am very contemptuous of such lazy middlebrow throwaway bullshit comments, and always have been (see my recent comments on DALL-E 2 responding to people who have not taken 5 seconds to google things, or Mark Friedenbach freely bullshitting left and right about crypto and supercomputers and anonymity). As I said then, I'm not upset that you disagree with me or are wrong; I wouldn't have added comments about China if I had thought that everyone agreed with me on it or that it was so obvious as to be beyond serious argument. I'm upset that you apparently don't care in the least bit and put zero effort into being less wrong, and could not be bothered to even hint at anything that could be considered a genuine argument or point.
Replies from: M. Y. Zuo↑ comment by lc · 2022-06-01T04:37:48.973Z · LW(p) · GW(p)
I think the argument about what kind of maturity and formality standards we should have on LW is much more complicated and not as cut and dry as you seem to think it is. I didn't have a negative reaction to his reply at all; it was funny and probably literally correct!
Replies from: M. Y. Zuo, Dirichlet-to-Neumann↑ comment by M. Y. Zuo · 2022-06-01T05:04:00.305Z · LW(p) · GW(p)
I’m not against biting one-liners, not even the occasional snarky zinger. The less respectful though, the more it’s inviting a race to the bottom.
Plus mimicry is a tactic usually used by throwaways or those trying to deflect from a flawed position, an emotional hangup, etc., so there’s a damned by association aspect as well, regardless of the writer’s intentions.
↑ comment by Dirichlet-to-Neumann · 2022-06-01T09:19:20.903Z · LW(p) · GW(p)
People with 60 000 karma should be much more careful about how they address people with 15 karma than the reverse.
Replies from: SaidAchmiz↑ comment by Said Achmiz (SaidAchmiz) · 2022-06-04T21:28:59.603Z · LW(p) · GW(p)
Are you suggesting that a high user karma value should impede one’s ability to speak straightforwardly and write concise, useful comments?
If so, then a high user karma value should be seen as an impediment, or a punishment. What, in your view, has gwern done to deserve such a handicap?
↑ comment by David Johnston (david-johnston) · 2022-06-02T03:05:28.841Z · LW(p) · GW(p)
For what it's worth, I agree that there's clear evidence of ill-will towards the Chinese government (and, you know, I don't like them either). It's reasonable to suspect that this might colour a person's perception of the state of thing that the Chinese government is involved with. It is also superficial, so it's not like I can draw any independent conclusions from it to defray suspicions of bias. I'm also not giving it a lot of weight.
2 comments
Comments sorted by top scores.
comment by dkirmani · 2022-05-31T11:10:46.162Z · LW(p) · GW(p)
Relatedly, what are the odds that Chinese capabilities advancements are kept secret?
Replies from: burmesetheaterwide↑ comment by burmesetheater (burmesetheaterwide) · 2022-05-31T19:47:45.811Z · LW(p) · GW(p)
Probability is high that all nations with strong AI research are keeping secrets, since some AI research will naturally go into projects with high secrecy. A better question is what the proportion of published to secret research is in USA, China, etc. It might actually be similar, which could suggest that China is pretty far behind.