ChinAI Newsletter #30: A Tale of Two Think Tanks and their Two Reports on AI

Welcome to the ChinAI Newsletter!

These are Jeff Ding's weekly translations of writings on AI policy and strategy from Chinese thinkers. I'll also include general links to all things at the intersection of China and AI. Please share the subscription link if you think this stuff is cool. Here's an archive of all past issues. *Subscribers are welcome to share excerpts from these translations as long as my original translation is cited.

I'm a grad student at the University of Oxford where I'm the China lead for the Governance of AI Program, Future of Humanity Institute.

CAICT Report: AI Development White Paper

First translation this week comes from the China Academy of Information Communications Technology (CAICT), a government think tank. You may remember them as the joint author of the Tencent book on AI strategy, which was the starting point of this newsletter. They’ve been around since 1957 and play a key role in shaping China’s IT policy. Thanks to Cameron Hickert, who contributed to an earlier issue of the newsletter on civil-military fusion, for his work on translating excerpts of the Technical Framework section of this week’s White Paper. 

These sections give a good look into how this quasi-government think tank views the current AI landscape, with some good technical details. The data issues section reflects a lot of the material from past newsletters: poor circulation of data, poor quality of data, and key dataset deficiencies. There’s an emphasis on the intermediate representation layer at the compiler level: basically, how do companies and engineers juggle multiple deep learning frameworks that run on different hardware platforms, and which layer will enable the best balance. Dive in for more below:

CAICT Report: AI Development White Paper — Technical Framework

More Historical Background on China’s AI + Security Industry

Last week, we looked closely at China’s AI + Security Industry in Xinjiang. This week’s report translation is from Yiou Intelligence, a think tank that operates completely differently from CAICT. It’s a private think tank that operates also as a “innovation service platform” that connects entrepreneurs, hosts summits, and gets funding from VCs. Definitely more fun reading reports from think tanks like Yiou - this one is a slide deck version that gives a rundown of some of the history of China’s AI + Security Industry.

Only did three slides so far, but they give a nice timeline of China’s AI + Security Industry which Yiou Intelligence argues dates back to 2011. Then, you had facial recognition companies like Yitu already cooperating with public security bureaus to identify cars with fake license plates. There’s also a close relationship to government plans, such as the Skynet and Sharp Eyes projects, as well as a national standard on video surveillance networking systems. For those interested in the broader view of AI + Security, see below:

Full Translation: 2018 China AI + Security Industry Development Research Report

This Week's ChinAI Links

Kai-fu Lee’s book AI Superpowers: China, Silicon Valley, and the New World Order is a must-read. His interview on Azeem Azhar’s Exponential View is a good listen.

Matt Sheehan, Kai-fu’s collaborator in writing the book, has a great Chinafornia newsletter that touches on the main themes of the book, particularly the strong connections between China and California’s tech scene.

Two reading recs from h/t Shazeda Ahmed:

The MIT Gender Shades project evaluates the accuracy of AI-powered gender classification product. You can see how Chinese startup Face++ compares with Microsoft and IBM.

This paper on the social credit system framed as part of a “reputation state” by Xin Dai, of Ocean University of China, is worth a read.

Thank you for reading and engaging.

Shout out to everyone who is commenting on the translations - idea is to build up a community of people interested in this stuff. You can contact me at or on Twitter at @jjding99