ChinAI #61: A Backlash to Social Credit Blacklists?

Plus, note collections on language asymmetry in AI + McGregor's book The Party

Welcome to the ChinAI Newsletter!

These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is a Rhodes Scholar at Oxford, PhD candidate in International Relations, Researcher at GovAI/Future of Humanity Institute, and Research Fellow at the Center for Security and Emerging Technology. Check out the archive of all past issues here and subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (those who can pay support access to all content for all).

Two quick plugs: going to be trying out public note-taking as a use-case for Twitter — see the two continuing threads below:

Feature Translation: Credit Blacklists, Not the Solution to Every Social Problem

This week’s translation is a joint work with Ryan Soh - a first-time contributor to ChinAI and an emerging stock market analyst who was previously a Chinese language student at Tsinghua University’s Inter-University Program. Please give him some love on Twitter (hyperlink in name above)!

We were tipped off to this piece by this August 1 issue of the excellent American Mandarin Society newsletter: “A July 30 opinion piece in the 钱江晚报, posted on people.cn, used a case in Shandong to push back on the idea of an all-encompassing social credit system. While there are many critics of the concept of such a system outside of China, little attention has been paid to discussions within China. This article is worth a read for how such a potentially sensitive topic can be approached within Chinese.” The case involved a teacher placed on a social credit blacklist for beating pupils with textbooks.

Ryan’s main points to emphasize: This article is interesting because it is in the Chinese public domain, and so up for grabs: a. Public discussion & input is seen as beneficial, rather than perfunctory. b. It indicates the issues are still being figured out. c. Therefore, rushing to judgement is presumptuous.

Shout-outs to Shazeda Ahmed (a PhD candidate at Berkeley and summer fellow at Upturn) and Jeremy Daum (a fellow at Yale Law’s China Center and creator of China Law Translate) for taking a look at the full translations. I’m trying to involve folks to play an “editing role” for each issue of the newsletter (I outlined why I think newsletters need editors in this refections segment from two weeks ago), and I can’t think of two better people for an issue on social credit blacklists. Jeremy shepherded us to the the second translated piece from The Paper (*both translations are included on the same Google doc this week) which contains more details on the implementation process of the blacklists; and an extra shout-out to Shazeda who helped us clarify terms of art in the translation and also was kind enough to do a full “sanity check” on my analysis — which follows:

Three main takeaways from my end:

1) I think many readers will be surprised to see such open criticism of the social credit system — both translations referenced broad public concern and controversy over the teacher being placed on the social credit blacklist. I think our collective surprise may be partly a product of the “techno-orientalist” lens through which we view China. The “techno-orientalist” frame dehumanizes Chinese people (Chinese people don’t care about privacy) and warps our understanding of China and Chinese people as vehicles where we project our own technological desires and anxieties. Xiaowei Wang, creative director at Logic Magazine, discusses this concept in the context of Jeremy’s work and Shazeda’s piece for Logic here.

2) There is a great deal of confusion and misunderstanding about the “social credit system” in China as well. Shazeda emphasizes that the consensus among law scholars in China is that social credit is meant to be a tool of administrative law, but this Qianjiang Evening News commentator thinks these will conflict, “What kind of mess would it be, if administrative and law enforcement offenses resulted in a loss to creditworthiness?” This confusion is understandable because “credit system” (信用体系) could refer to the national or provincial level social credit systems (administrative law vehicles), which differs from "People’s Bank of China’s credit information evaluation system, blacklists for those deemed to be untrustworthy by the People’s Court, etc. The second article in the full translation highlights these different systems.

3) These articles are notably tech-free. There’s nothing about using a bajillion parameters to calculate someone’s “social credit score.” Shazeda comments, “I've been cautious about how I talk about technology in relation to the social credit system because even though I've spotted a few vague, aspirational policy references to "implement big data and AI into the credit system," thus far the applications I've seen are limited and fragmented.” A related takeaway is that this is a nascent and evolving implementation process, as emphasized at the end of the article from The Paper.

SEE BOTH FULL TRANSLATIONS: Qianjiang Evening News: Credit Blacklists, Not the Solution to Every Social Problem; and The Paper: Wulian County Bureau of Education and Sports further explain recent teacher sanctions

ChinAI Links (Four to Forward)

We have to start understanding China’s tech scene in context that involves countries other than the U.S. - one platform consistently doing an awesome job at covering this broader landscape is FT’s Tech Scroll Asia, edited by James Kynge. Check out this recent edition on the Japan-Korea tech row that’s threatening supply chains.

I’ve been consistently impressed by FT’s China+tech reporting, and I think they have the best team in the game, so here’s a second FT link: Yuan Yang, a China tech correspondent, has a column for FT magazine that is consistently sublime. The most recent one recounted her experience getting her face scanned in Xinjiang and also included details like “among the Han not everyone felt as warmly about the cameras. A shopkeeper cited them as one reason his clientele had been dwindling – the ever-present security on the streets had made it more inconvenient for people to come out, he said.” Also, see a previous column on the Anti-996 campaign that stirred China’s tech world, which included some cool anecdotes from her friends about absurd work practices.

In MacroPolo’s continuing series on China’s AI talent, Joy Dantong Ma takes an interesting cross-section of the global AI talent pool (authors who have had papers accepted to the NeurIPS (previously known as NIPS) conference. One main finding: “When it comes to retention, of the 2,800 Chinese accepted to NeurIPS over the last decade, about three-quarters of them (>2,000) are currently working outside of China, according to my estimates (see Figure 2). And of the 2,000+ AI talent that have left China, about 1,700 (85%) came to the United States.” While I think this finding largely makes sense, I would like to have seen more explanation of the methodology (Note: Joy said, “The approach I took was indeed sampling. There were over 15,000 authors over the past decade so it wasn’t within my resources to track down all of them. After getting the sample…what I did was to analyze each location individually in a hypergeometric distribution so it is more robust”). I would also be curious if this trend has shifted in the past couple years.

Of course it’s the smartest VCs that will be among the first to take advantage of the enormous "information/knowledge arbitrage” of translating Chinese S&T media, which ChinAI is built on. See these excellent pieces by Andreesen Horowitz’s team (Connie Chan with help from Avery Segal) that feature translated excerpts — one from a speech given by the “father of Wechat” and one of field notes from an interview by Ren Zhengfei, Huawei’s CEO, given to Chinese media in Shenzhen.