A growing number of AI research papers published at the prestigious Neural Information Processing Systems (NeurIPS) conference reveal a surprising level of collaboration between researchers from the US and China, suggesting that despite their rivalry in the field, both countries are heavily relying on each other to advance their artificial intelligence capabilities.
According to an analysis conducted by WIRED using OpenAI's Codex tool, over 3% of the total papers presented at NeurIPS last year involved collaboration between authors from US and Chinese institutions. This finding holds true for a recent count as well, with roughly 3% of the papers involving both sides in 2024. Furthermore, research on transformer architecture developed by Google researchers has been widely adopted in China, appearing in almost 30% of all papers featuring Chinese authors.
Moreover, Meta's large language model Llama was prominently featured in over a fifth of papers co-authored with Chinese institutions and is seen in more than 60 papers from US-based researchers. The trend holds for other key models like the Qwen large language model developed by Alibaba, appearing in nearly two-thirds of research papers co-authored with authors from the US.
Dr Jeffrey Ding, an expert on China's AI landscape, notes that "the collaboration between the US and China is not surprising" given their intertwined AI ecosystems. He further states that policymakers from both countries would benefit from acknowledging this cooperation rather than downplaying it.
Some researchers emphasize the role of personal relationships and international collaborations in fostering a culture of knowledge sharing. Dr Katherine Gorman of NeurIPS highlights how "NeurIPS itself is an example of international collaboration" as well as the long-lasting bonds formed between colleagues, which persist even after students leave their universities.
Despite fears from US politicians and tech executives about China's influence on AI research, this study serves as a reminder that there are many benefits for both countries to engage in collaborative research.
According to an analysis conducted by WIRED using OpenAI's Codex tool, over 3% of the total papers presented at NeurIPS last year involved collaboration between authors from US and Chinese institutions. This finding holds true for a recent count as well, with roughly 3% of the papers involving both sides in 2024. Furthermore, research on transformer architecture developed by Google researchers has been widely adopted in China, appearing in almost 30% of all papers featuring Chinese authors.
Moreover, Meta's large language model Llama was prominently featured in over a fifth of papers co-authored with Chinese institutions and is seen in more than 60 papers from US-based researchers. The trend holds for other key models like the Qwen large language model developed by Alibaba, appearing in nearly two-thirds of research papers co-authored with authors from the US.
Dr Jeffrey Ding, an expert on China's AI landscape, notes that "the collaboration between the US and China is not surprising" given their intertwined AI ecosystems. He further states that policymakers from both countries would benefit from acknowledging this cooperation rather than downplaying it.
Some researchers emphasize the role of personal relationships and international collaborations in fostering a culture of knowledge sharing. Dr Katherine Gorman of NeurIPS highlights how "NeurIPS itself is an example of international collaboration" as well as the long-lasting bonds formed between colleagues, which persist even after students leave their universities.
Despite fears from US politicians and tech executives about China's influence on AI research, this study serves as a reminder that there are many benefits for both countries to engage in collaborative research.
. It's like in life, we often need others' perspectives to see things more clearly and come up with better solutions 
. This AI thing is no exception. Just because they're competing doesn't mean they can't work together and learn from each other 
. Maybe it's time for us all to take a chill pill and appreciate the beauty of collaboration, rather than trying to one-up each other 
.
️ and it totally made me think about how hard it is to work on complex AI projects without anyone to chat with. I mean, some of these researchers have been working non-stop for months, and they still need to collaborate with others? That's like trying to make a video game solo – it just wouldn't be the same 
.
I've been going there every morning to get my fix. But anyway, back to AI... what do you think about all these researchers collaborating with each other? Do you think it'll lead to any breakthroughs or just more coffee breaks? 
They'd be like, totally stuck in their own little bubbles, not learning anything from each other 
And I love how Dr Gorman said NeurIPS is an example of international collaboration - that's what it's all about! 
like they're secretly BFFs or something lol. You'd think with all the drama between them, it'd be hard to get anything done together... but I guess that's just how the tech cookie crumbles 
i mean, its not all about the rivalry, ppl r actually helpin each other out 
. its like a big ol' knowledge sharing fest 
! and dr ding says its all about their ecosystems bein intertwined 
. just think of it as international collaboration 
...
Maybe this is what 'global governance' looks like? 
what's scary for me tho is when ppl start hyping up "us vs china" in AI research 
...
. It's a win-win! Moreover, these collaborations aren't just about sharing knowledge; they're also about fostering relationships and building trust 
. We're all in this together, after all 
.
we should be focusing on how our global collaborations can drive real progress 
I know we're all about collaboration and sharing knowledge, but can't we get some better display options than this? 
. And what's up with Meta and Google basically owning the Chinese AI scene 
?
. Plus, it's great to see NeurIPS itself as an example of global collaboration 
. We should be celebrating this kind of cooperation rather than fearing it 
. Like, what about when you're working on a project with someone from another part of the world, but you can't even access their actual code or data without going through some kind of VPN hell 
. Like, are there any researchers who aren't participating in these international projects? It's all pretty fascinating stuff, but also a bit unsettling when you start thinking about the bigger picture