Back in 2017 I started wondering what 🇨🇳 scientists & policymakers thought about AI safety. I asked around, didn’t find much. I’ve followed this closely since but been skeptical it’s getting real attention.

But in the past 18 months that’s changed. 🧵
carnegieendowment.org/research/2024/… 2/ Few years ago it was rare to see Chinese research / public discussion of AI safety. There were exceptions, w/ influential scientists like Gao Wen talking about AGI risks, & Ministry of S&T docs referencing "ensuring human control over AI."

But these were far from mainstream.

Tomorrow 🇨🇳 will attend the UK AI Safety Summit.

But “China” or even “the Chinese government” isn’t one thing. Several different ministries are angling to lead on Chinese AI governance at home & internationally.

Short 🧵 on what we do/don’t know, about those bureaucracies: 1st, to dispel common myth: Not all 🇨🇳 AI governance comes directly from Xi Jinping. Xi gives high-level guidance & nothing happens if he explicitly opposes it.

But Chinese AI policy shaped by tons of policy actors w/ own concerns+agendas. Report on this:
carnegieendowment.org/2023/07/10/chi…

AI Red Teaming w/ 🇨🇳 Characteristics

A key Chinese standards body released a draft standard on how to comply w/ China's generative AI regulation. It tells companies how to red team their models for illegal or "unhealthy" information.

🧵 on a fascinating document: First, the context:
China has been rolling out regulations on algorithms & AI for ~2 years, including a July regulation on generative AI.

All these regs focus on AI's role in generating / disseminating info online. More background below.
2/x
carnegieendowment.org/2023/07/10/chi…

📢China just released its much-anticipated regulation on generative AI.

🧵below w/ my initial notes/reactions.

TLDR: the final version is *much* less strict than the April draft version. This reflects a very active policy debate in 🇨🇳 + econ concerns.
https://t.co/fblFYZ5bVbcac.gov.cn/2023-07/13/c_1…
2/x
Scope: Article 2 dramatically narrows the scope of AIGC activities covered. The draft version included R&D, general use of AIGC, and provision of AIGC services to the public.

This version cuts R&D and general use, and covers just providing AIGC to public. *Big* change

I’ve spent the 1.5 years trying to answer two questions:
1. What is China doing on AI governance?
2. Who are the key players, and how do 🇨🇳 AI regulations actually get made?

Today we’re publishing a paper trying to answer that. 🧵 w/ key takeaways (1/10)
https://t.co/yBuHWLfdiCcarnegieendowment.org/2023/07/10/chi…
China's AI regs will shape the technology’s trajectory in China & around 🌍.

They'll be “default settings” for their AI exports (LLMs in Indonesia, AVs everywhere). As 🇺🇸+🇨🇳 systems interact more we need to understand each system's safeguards.

Bonus: we can learn sth. (2/10)

I strongly recommend reading this intro to generative AI regulation from @AlexCEngler.

I'm going to share a couple highlights + thoughts below, and would love to hear from others on this. Mini-🧵
brookings.edu/blog/techtank/… Impt section on AI value chain, and the different capabilities + responsibilities of upstream vs downstream developers. This is complicated & will crop up in more and more AI regulatory issues.

Alex makes his argument for 'making the model move' here: techpolicy.press/to-regulate-ge…

How did China go from a technological backwater to an innovation powerhouse in just 20 years?

It's a big question with big implications for 🇨🇳, 🇺🇸, and 🌍. I take a crack at answering it in my new piece for @ForeignAffairs.
🧵 1/x
foreignaffairs.com/articles/china… 2/x
First, what the piece is not doing:
It's not providing the be-all-end-all explanation that covers all variables. I had 2,000 words. 🤷‍♂️

What it is trying to do:
Provide a coherent framework for understanding the main policy drivers of China's innovation boom.

🧵 China has been rolling out lots of new AI governance measures that will reshape 🇨🇳 AI & make waves around the 🌏.

It’s easy to lose the governance forest for policy trees, so I broke down 3 key initiatives, showing how they'll compete & converge. (1/x)
carnegieendowment.org/2022/01/04/chi… China’s widespread abuse of AI surveillance tech leads many to dismiss any governance moves there. But that risks being a huge strategic oversight.

Anyone who wants to compete against, cooperate with, or simply understand Chinese AI must examine these moves closely. (2/x)

THREAD: This yr I got a lot of people asking “How does Silicon Valley see China? How has that changed?”

People are usually looking for a 1-dimensional answer (partner➡️rival!) but SV-🇨🇳 ties are way more tangled up than that.

So I made a chart! (1/x) macropolo.org/how-silicon-va… Impt qualifier: this chart is built off my *subjective assessment* of these trends.

That assessment is built on >400 interviews, off-record convos, & projects done w/ Silicon Valley investors, entrepreneurs, researchers, etc.

But it's still *subjective* so take it as such.

THREAD: Today our @macropolochina team dropped a new report forecasting Chinese politics, economics, technology & energy 2020-2025.

I took on the task of predicting what will go down in Chinese tech over the next 5 years.

Here’s what I came up with: macropolo.org/analysis/china… First, a limitation: I didn’t try to cover every tech sector+issue. (Quantum, surveillance, fintech, social credit, etc.) Imagine predicting everything in US tech 2015-2020...

So if I didn’t cover your area, you should write that piece. Send it to me, I’ll read it! (thread 2/x)

Key dimension of US-China tech competition is currently playing out in countries across the developing world.

So I analyzed data on the most downloaded apps in six key countries for 2015 & 2019. Here’s what I found:

thread 1/ US apps remained dominant in 2019 for most markets, often w/ 60-75% market share among the most downloaded apps.

The one big exception here is India, which we'll dig into below... 2/

🎙️New episode of Heartland Mainland: the Iowa China Podcast!🎙️

For the past yr, @JianingHollyHe & I have been talking w/ Rick Kimberley: Iowa farmer ➡️ ag celebrity in China ➡️ collateral damage in the trade war.
Here's what we learned. /thread
macropolo.org/podcast/heartl… @JianingHollyHe We spoke w/ a soybean industry rep who told us they first set up a Beijing office in 1982, betting that Reform + Opening would mean greater demand for animal feed & soybeans.

That big bet on Chinese markets didn't start paying off for 2 decades, until China entered WTO in 2001.