Great article as always. I was excited to share this with my american friends.
> "Almost every team we sat down with admitted, in some form, that they were scared of Doubao"
0) Truly understated by X(twitter) how bytedance is closest to Google/Alphabet in china. Doubao is actually very pleasant to use better than chatgpt and gemini.
Chinese ecosystem has always leaned more into voice interface and it's maturing well.
1) doubao's frontier model (`doubao-seed-2-0-pro-260215`) isn't even listed on any of benchmark sites like Artificial Analysis, lmarena, etc.
Likely because it's only(?) available through their cloud service ('volcengine') and not available through aggregators like OpenRouter.
But good example of tremendous dichotomy b/w how much china is up to date with developments in usa verses vice versa.
Great question to surface, and the 'state defines, industry executes' framing is sharp! One more dimension I'd be curious to hear your take on: does the relative weight Chinese education places on STEM over humanities/liberal arts also shape who ends up self-selecting into AI research, and what meta-questions they're capable (or not capable) to ask? (I sense similar dynamic in Silicon Valley too, so maybe it's not purely a China story.)
Really interesting article, thank you for writing and sharing. One of the things I find most interesting is the social contract of AI and the difference between who is allowed to drive the future (and/or at least take credit for it). My question is...what is the opportunity for the people to determine the future outcomes and define the social contract in the US, and how that narrative would influence the narratives and state mandate in China? You mentioned briefly that the Silicon Valley ethos is driven by market viability and sentiment, but for several reasons this could just be another situation of the tail wagging the dog (I believe you may have alluded to this as well). I'd like to hear your thoughts on what is seemingly the current trajectories, and if there exists alternatives to this trajectory to fight the issues you mentioned (AI and inequality, joblessness, human meaning). Thanks again
I'd be curious to hear if any of the Chinese companies said (or were allowed to say) anything about the massive negative environmental impacts of their work. Is it currently the same as all the American companies where they either never mention it, or at most, wave their hands away and say that AI will magically solve that too?
A far higher portion of Chinas energy compared to Americas is already renewable. About 85% IIRC, and obviously they lead the world in terms if the development and production of both electrovoltaics and wind turbines. There's obviously the other factors like water use & critical mineral mining (which they extract themselves anyway) but on the whole it still seems like they've got a better handle it the either the US or Europe does.
Here is a another view of the same dataset that article is referencing (ember-energy). According to the same source the IER quoted, the US produces 25.6 percent of its electricity from renewables, while China is at 37 percent. The IER article also conflated energy production with electrical generation. It's all quite misleading.
A report from an American based explicitly pro-neoliberal energy market think tank is about as biased a source as you can get when discussing China (and extremely so). It's like expecting the Times of Israel to give an objective account of the ethnic cleansing Israel is doing to Palestinians.
So it is! I just did a quick search. But it's consistent with all reports I've read in the recent past (and ditto for India). Lots of investment in green stuff, but also tons of investment in coal plants. That squares with my impressions from living in China and people I know still there. And of course, the US gov is doing everything possible at the moment to revert back to the Industrial Revolution/destroy all parks and funding for solar, hydro, wind etc. Fun times!
Afra: as for your Giorgio de Chirico-like description of Xiong’an at the end of the piece, one is reminded of Xu Zhiyong’s powerful critique of Xi Jinping’s hi-tech folly:
Great article, very timely. Let's see what comes (or doesn't) after the meeting between heads of state this week. I read some Wall Street thing talking about trading magnets for lithography gear... Marco Polo would have been impressed
Those in Silicon Valley are allowed to have gods before the state. Some make themselves the gods, creating the future. Some make tech progress the future, humbly following its divination. In China there’s no space for considering what is driving the future outside of the state
Great article as always. I was excited to share this with my american friends.
> "Almost every team we sat down with admitted, in some form, that they were scared of Doubao"
0) Truly understated by X(twitter) how bytedance is closest to Google/Alphabet in china. Doubao is actually very pleasant to use better than chatgpt and gemini.
Chinese ecosystem has always leaned more into voice interface and it's maturing well.
1) doubao's frontier model (`doubao-seed-2-0-pro-260215`) isn't even listed on any of benchmark sites like Artificial Analysis, lmarena, etc.
Likely because it's only(?) available through their cloud service ('volcengine') and not available through aggregators like OpenRouter.
But good example of tremendous dichotomy b/w how much china is up to date with developments in usa verses vice versa.
also i feel like doubao isn’t trying to be the “most capable" model” but the “most useful tool”
Frontier culture contrasted with the Celestial Empire…
Interesting
This has happened before. 🫡
Great question to surface, and the 'state defines, industry executes' framing is sharp! One more dimension I'd be curious to hear your take on: does the relative weight Chinese education places on STEM over humanities/liberal arts also shape who ends up self-selecting into AI research, and what meta-questions they're capable (or not capable) to ask? (I sense similar dynamic in Silicon Valley too, so maybe it's not purely a China story.)
Really interesting article, thank you for writing and sharing. One of the things I find most interesting is the social contract of AI and the difference between who is allowed to drive the future (and/or at least take credit for it). My question is...what is the opportunity for the people to determine the future outcomes and define the social contract in the US, and how that narrative would influence the narratives and state mandate in China? You mentioned briefly that the Silicon Valley ethos is driven by market viability and sentiment, but for several reasons this could just be another situation of the tail wagging the dog (I believe you may have alluded to this as well). I'd like to hear your thoughts on what is seemingly the current trajectories, and if there exists alternatives to this trajectory to fight the issues you mentioned (AI and inequality, joblessness, human meaning). Thanks again
I'd be curious to hear if any of the Chinese companies said (or were allowed to say) anything about the massive negative environmental impacts of their work. Is it currently the same as all the American companies where they either never mention it, or at most, wave their hands away and say that AI will magically solve that too?
A far higher portion of Chinas energy compared to Americas is already renewable. About 85% IIRC, and obviously they lead the world in terms if the development and production of both electrovoltaics and wind turbines. There's obviously the other factors like water use & critical mineral mining (which they extract themselves anyway) but on the whole it still seems like they've got a better handle it the either the US or Europe does.
Nope: https://www.instituteforenergyresearch.org/fossil-fuels/china-touts-its-renewables-but-fossil-fuels-rule/
Here is a another view of the same dataset that article is referencing (ember-energy). According to the same source the IER quoted, the US produces 25.6 percent of its electricity from renewables, while China is at 37 percent. The IER article also conflated energy production with electrical generation. It's all quite misleading.
A report from an American based explicitly pro-neoliberal energy market think tank is about as biased a source as you can get when discussing China (and extremely so). It's like expecting the Times of Israel to give an objective account of the ethnic cleansing Israel is doing to Palestinians.
So it is! I just did a quick search. But it's consistent with all reports I've read in the recent past (and ditto for India). Lots of investment in green stuff, but also tons of investment in coal plants. That squares with my impressions from living in China and people I know still there. And of course, the US gov is doing everything possible at the moment to revert back to the Industrial Revolution/destroy all parks and funding for solar, hydro, wind etc. Fun times!
Afra: as for your Giorgio de Chirico-like description of Xiong’an at the end of the piece, one is reminded of Xu Zhiyong’s powerful critique of Xi Jinping’s hi-tech folly:
https://chinaheritage.net/journal/where-is-chinas-intelligentsia-during-the-covid-emergency-re-reading-xu-zhiyongs-letter-to-xi-jinping/
LLMs are consumer products. They’re a hobby compared to their industrial counterparts
Great article, very timely. Let's see what comes (or doesn't) after the meeting between heads of state this week. I read some Wall Street thing talking about trading magnets for lithography gear... Marco Polo would have been impressed
Those in Silicon Valley are allowed to have gods before the state. Some make themselves the gods, creating the future. Some make tech progress the future, humbly following its divination. In China there’s no space for considering what is driving the future outside of the state
thank u will fix