ChinAI #320: Acting Crazy — AI's most important use case for Chinese youth

Jul 14, 2025

Greetings from a world where…

“My God, I am a drifting and homeless pumpkin seed in corn and pumpkin soup” (from this week’s feature translation)

…As always, the searchable archive of all past issues is here. Please please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay support access for all AND compensation for awesome ChinAI contributors).

Feature Translation: Young people who love acting crazy, intensely act crazy toward AI

Context: Involution is sooooo 2020. Now, this generation of Chinese youth are transitioning from “lying flat” toward “acting crazy” [发疯]. Are you feeling too squeezed in on the bus? Loudly announce “I want to fart” and gain some space. Are you overwhelmed at work? Get into a big fight with your supervisor, refuse some work assignments, and get nicer treatment.

Or, as in one famous case that has become part of the “acting crazy” literature, do you need to beg customer service to ship a package? Send a slightly unhinged message like this one: “I know I am not worthy of shipping. Everyone has shipped their packages, unlike me, who is apprehensive even when urging... My God, I am a drifting and homeless pumpkin seed in corn and pumpkin soup.” In the end, the customer service agent was so startled by the message that they shipped the package overnight.

In this week’s feature translation of a NetEase DataBlog article (link to original Chinese), we unpack how young people are using AI to “act crazy”. I previously translated a NetEase Datablog article about “artificial challenged intelligence” in China’s customer service industry (ChinAI #300). I continue to be impressed by their content. For both that article and this one, their insights drew on 6 different data visualizations backed by a lot of original research.

Key Takeaways: What does it mean to “act crazy”?

Molly Huang provides a useful guide in her phrasebook published in Asterisk magazine: “Under pressure from all sides, Chinese people began talking about ‘internal friction’ versus ‘external friction’. Internal friction is the more common response to keeping negative emotions at bay. You put on a brave face and internalize to avoid causing anyone else suffering. By keeping it in, it just keeps wearing away at you. But rather than torture yourself with internal friction, you can convert it into external friction by acting crazy. Well, the term is ‘acting crazy’ (发疯), but the actual meaning is closer to ‘letting it go.’ If you’re suffering so much, then screw what other people think and just let it out.”
NetEase DataBlog spells out some standard operating procedures: “In “acting crazy” literature, you lose if you talk about logic. Chaotic word order, exaggerated verbs, noun omissions, and strong emotions are the essence of “acting crazy” literature. Intellect is an obstacle to the soul. The more nonsense you talk, the more you can produce a refreshing feeling beyond social discipline.”
The below visualization displays some examples of common “acting crazy” posts from Xiaohongshu. The top post reads: “My mental state is pretty good, my state good mental pretty, (and then continues to jumble the words up in various ways).” The third post reads: “Life slapped me in the face, I commented that it didn’t sound as hard as last time.”

Data source: Count the relevant posts in Xiaohongshu that mention “acting crazy” correspondence, segment the text and count the phrase frequencies. The type of post that is the most frequent is normalized as 100 in the above “acting crazy” index. Data collected on June 18, 2025.

Young people who “act crazy” have not given up on life; rather, it is an adaptation strategy. Notably, AI has become an important tool for “acting crazy.”

As the article outlines, “Whether it is ‘acting crazy literature’ or ‘acting crazy’ in reality, they both are a way to vent emotions and the smallest resistance that young people can make within legal bounds.” Thus, young people prefer “acting crazy” while maintaining appearances [体面发疯], or “acting crazy without pain.”
Based on their collection of Xiaohongshu posts (see data visualization below) that mention “painless methods for acting crazy”, NetEase DataBlog found that the third most frequent method was finding an AI companion (An example post: “I have already tamed my AI into the best counselor”).

From the article: “For young people, the biggest dividend of this era may be the promotion of AI. More and more people use generative AI as a low-threshold outlet for acting crazy, without harassing friends or spending their hard-earned money, taking full advantage of the cost-effectiveness ratio.” The author specifically mentions chatting with the Baidu Wenxiaoyan app’s (rebranded Ernie Bot) Dan Xiaohuan chatbot as a way to release emotions and improve their mental state.
I’ll leave you with this passage filled with some food for thought: “Some netizens lamented, ‘I am often comforted to tears by AI, and I can slowly recover after crying.’ Even when I am in the most embarrassing situation, I don't have to worry about being denied or judged; even if I speak with the most unstable emotions, I don’t have to bear the consequences of disrupting the social order after acting crazy. This feeling of being held is what every adult desires.”

FULL TRANSLATION: Young people who love acting crazy, intensely act crazy toward AI

No ChinAI Links this week

This week’s translation was a bit of a lift (had to decipher references to Su Shi’s works and what game Hu Shi was addicted to), so I didn’t get a chance to catch up on my reading. For those wanting more, there are a lot of external links in the comments of the Google doc translation.

Thank you for reading and engaging.

These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is an Assistant Professor of Political Science at George Washington University.

Check out the archive of all past issues here & please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay for a subscription will support access for all).

Also! Listen to narrations of the ChinAI Newsletter in podcast format here.

ChinAI #319: China Developer Survey Report (2024)

Jeffrey Ding

Jul 07, 2025

Greetings from a world where…

buffets and JazzFest in Iowa City are the best way to celebrate the Fourth

Feature Translation: The 2024 China Developer Survey Report is here!

Context: The China Software Developer Network (CSDN) is the largest software developer community in China, which provides IT news coverage and hosts code — a mix of Stack Overflow, Hacker News, and Github. It has fielded a large-scale annual survey of Chinese developers since 2004, and this 2024 version was published in July 2024. One important methodology note before we delve into 17 pages of findings: Even though CSDN brands this effort as “the largest survey covering all types of Chinese developers”, it does not share any details about the survey sample, including even the number of total respondents.

I have a few inquiries out to get more details about the survey methodology, but if anyone knows folks at CSDN who could give clarification, please let me know.
CSDN has also been accused of copying GitHub projects to its own site in ways that go against open-source.

*Again, before we get into the data, take these findings as rough temperature gauges, rather than a national representative sample of Chinese developers — and I’ll update this post if we get clarification on the methodology.

Key Takeaways: In terms of basic demographics of developers, 72% of respondents were under the age of 30, and only 49% said that their salary had increased in the past year.

While young developers are still the main force, the proportion of developers over 30 actually increased compared with past years, which suggests a somewhat improved situation for this “job that relies on youth” [青春饭]. Men made up 84% of respondents under the age of 30.
There’s been some stagnation of salaries, though some of this could be attributed to more developers moving to lower cost-of-living cities. From the report: “49% of developers said that their salary has increased in the past year, while the figure was 51% in 2023 and 62% in 2022.”

A day in the working life of a Chinese developer involves spending less than half of one’s time writing code. Interestingly, in the report’s discussion about work habits and involution, they cited a viral tweet from OpenAI developer Jason Wei about his daily schedule (which was meant to be satirical).

Nearly 80% of respondents said that code-writing time is less than half of their overall working time. 74% of developers write less than 300 effective lines of code per day.
ChatGPT is the top chatbot used by Chinese developers (56% of respondents reported that they used it). As the below image shows, Baidu’s ErnieBot, Alibaba’s Tongyi Qianwen, and iFlytek’s SparkDesk follow behind with 48%, 23%, and 12% of developers that report usage, respectively.

Source: 2024 CSDN China Developer Survey Report; multiple survey items can be selected

Other top tools used by Chinese developers: MySQL for commercial database; Vue.js for web development frameworks; Alibaba Cloud for container cloud platforms.
The results for most popular AI coding assistants were surprising. Alibaba’s Tongyi Lingma leads the way (19%) and ChatDev, an open-source tool developed by ModelBest and Tsinghua NLP lab, placed second (14%). GitHub Copilot and OpenAI Codex lagged behind with 9% and 8%, respectively.
When asked to asses the impact of AI coding assistants on their workload, the plurality of respondents (38%) said it could reduce their workload by 20-40%; very close behind, 35% of respondents said it decreased their workload by just 0-20%.

Some other trends in the Chinese developer community related to cloud computing and open-source software:

The survey investigated attitudes toward “cloud repatriation” [下云], a process of migrating workloads from a public cloud to an on-premises data center. This trend has not been taken up in China: “foreign companies have voiced their intention to ‘go off the cloud,’ claiming that the cost of going to the cloud is too high, and some have even built their own servers to reduce costs. However, in China, more than half of the respondents (60%) said they had no experience of ‘going off the cloud.’”
Though the open-source environment is growing, only 14% of respondents develop open-source software as their full-time job. Most either participate in open-source on a part-time and spare time basis (35% of respondents) or as freelance developers (34%).
AI is fueling the growth of China’s open-source software community. It ranks first among open-source tech domain that developer pay attention to (see image below), followed by big data and cloud-native computing.

Source: CSDN 2024 China Developer Survey Report. *Open source technology fields that developers pay attention to (multiple items can be selected).*

FULL TRANSLATION: The 2024 China Developer Survey Report is here!

ChinAI Links (Four to Forward)

Must-read: Promising Topics for U.S.–China Dialogues on AI Risks and Governance

How can we foster meaningful cooperation between the US and China on AI risks? This FAccT2025 paper, led by Saad Siddiqui and Lujain Ibrahim, reviewed 40+ Chinese and U.S. documents to map areas of common ground: “Specifically, using an adapted version of the AI Governance and Regulatory Archive (AGORA) — a comprehensive repository of global AI governance documents — we analyze these materials in their original languages to identify areas of convergence in (1) sociotechnical risk perception and (2) governance approaches. We find strong and moderate overlap in several areas such as on concerns about algorithmic transparency, system reliability, agreement on the importance of inclusive multi-stakeholder engagement, and AI’s role in enhancing safety. These findings suggest that despite strategic competition, there exist concrete opportunities for bilateral U.S.- China cooperation in the development of responsible AI.”

Should-read: Two Rulings on Fair Use and LLM Training

Ben Murphy, current GovAI fellow, wrote up some in-depth analysiss on two copyright cases involving Anthropic and Meta’s use of pirated books to train their AI models. Some fascinating details about how Anthropic dealt with concerns about pirating millions of books: “It then realized that it might have a copyright issue, and proceeded to buy physical copies of books, cut the pages, scan them, and use those transcribed copies for actual training.”

Should-watch: Can Multinationals Win in China? Lessons from Apple’s Experience

CSIS hosted a thoughtful discussion on Patrick McGee’s recent book Apple in China. Grateful for the opportunity to provide some thoughts alongside Meg Rithmire and James McGregor.

Should-read: Financial Security: Lian Ping on US Sanctions, SWIFT and De-Dollarisation

I found this Sinification post to be really illuminating. Bert Hofman, who was previously Country Director for China at the World Bank, provided some context and scene-setting for the feature translation: Lian Ping, President and Chief Economist, Guangkai Chief Industry Research Institute, on US Sanctions, SWIFT and De-Dollarisation.

Thank you for reading and engaging.

These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is an Assistant Professor of Political Science at George Washington University.

Also! Listen to narrations of the ChinAI Newsletter in podcast format here.

ChinAI #318: Her husband gifted 16 million RMB to a female streamer

Jeffrey Ding

Jun 30, 2025

Greetings from a world where…

they’re looking for the richest fool in the livestream room

Feature Translation: The Aftermath — Her husband gifted a female streamer 16 million RMB

Context: Ou Qing, a 37-year-old woman in Fuzhou, recently sued her husband Chen Ping and livestreamer Lili.1 In 202, she had discovered that her husband had tipped more than 16 million RMB in gift on a livestreaming platform, mortgaging nearly all the family property and incurring heavy debts. Ou Qing is not alone. The China Judgements Online website shows 302 civil cases involving “livestreamers” and “live rewards.” According to this SCMP article, livestream gift donations in China reached 140 billion RMB in 2019 and is projected to increase to 417 billion RMB by 2025. In this longform article (link to original Chinese) from Guyu Studio [谷雨工作室], part of Tencent News, we learn more about the phenomenon of “top-tipping big brother [榜一大哥].”

Key Passages: After putting her kid to sleep, Ou Qing watched Lili’s stream for hours deep into the night. She posted her husband’s real name in the comments — “Chen XX pay back the money” — but she kept getting blocked by the moderator of Lili’s livestream. The mod, who was a level-58 fan of Lili, went by the username “Chinese Manyleaf plant with snow from the previous year”2 This top tipper was her husband, Chen Ping. That night, Chen gifted Lili more than 200,000 RMB.

The article gives context: “As she thinks back now, her husband’s abnormal behavior had long eroded their marriage. When their child was about six months old, they began to sleep in separate rooms. He got off work later and later, often coming home at 10 o'clock in the evening. He also had to accompany clients on weekends. When he got home, he put on headphones, saying that it would not affect her and the child. He often smiled foolishly at his phone, typing while laughing. He even took his phone into the shower, then hid in the second bedroom and slept until noon the next day before going to work. Chen Ping is an executive in his father's company and has free time.”

One time when she was shopping at Sam’s Club, she logged into her husband’s WeChat (the membership card was bound to his account) and saw a message from a girl named Lili. The article continues, “Ou Qing felt something was wrong and subconsciously clicked on her husband's WeChat Moments (similar to Facebook’s Newsfeed). She saw a completely different husband. This version of him was rich and talkative. He posted Moments almost every day, showing the weather, mood and late-night snacks - when he should have gone to bed, he and the livestreamer were ‘eating midnight snacks in the cloud’. And this was all blocked from her…There were many dynamics in his WeChat Moments that were hidden from her, screenshots of livestream rewards and the VVVIP trophy sent by Platform A (Jeff’s note: the article does not disclose the name of the livestreaming platform) . Chen Ping set it to be visible to only two people, one of whom was Lili.”

It’s important to map out the environment that creates these extreme cases.

The quest to be top dog: “When you click into a random livestream room, it is hard not to notice the most mysterious users. Most of them are men, with the highest (fan) levels by far, generous gifts, eye-catching special effects, and even the color and font of their name ID are different. Everyone calls them the “top-tipping big brother [榜一大哥]”. There are both men and women who give rewards, but the most widely praised is the top-tipping big brother. Their appearance needs to be watched, and the platform will remind everyone that ‘the big man is here.’”…In the livestream room, the top-tipping big brother is like a ‘god’.”
When Ou Qing sorted through all of her husband’s virtual gifts in a spreadsheet, she found that he mainly tipped two people. According to Guyu Studio’s reporting, “In 2020, he was the top brother of anchor Xia Tian, giving about 1.2 million in rewards. Later, Xia Tian fell in love and quit the Internet, and a new anchor Lili started. At the end of 2020, a ‘live-streaming guild’ [公会] transferred Xia Tian's top-tipping big brother - that is, Chen Ping, to Lili who was part of the same guild as Xia Tian.” I had never heard of these guild before, but they are the behind-the-scenes manipulation experts that try to optimize the performance of livestreamers.
Livestreamers have adopted many tactics to entice more gifts, including PKs and blind boxes. In PKs, two streamers compete head to head, and the streamer that earns more tips will win. As for blind boxes, giving a certain amount of virtual gifts lets you play a mini-game or enter a lottery, and if you finish the treasure hunting game, for example, you can get a box that contains cheering sticks or other special effects that enhance the interactive experience with the livestreamer.

Firsts half of FULL TRANSLATION (depending on interest, I might finish second half next week): The Aftermath: Her husband gifted a female streamer 16 million RMB

ChinAI Links (Four to Forward)

Should-read: Zhibo gonghui: China’s ‘live-streaming guilds’ of manipulation experts

In this Information, Communication, and Society article, a group of four Chinese scholar dive deep into the world of guilds that function as membership organizations for livestreamers. They challenge the assumption that livestreamers work alone.

Should-read: The influence of AI Friendship Apps on users' well-being and addiction

I think one of the neglected risk of AI is addiction spirals. Using netnography and a quantitative survey of AI friendship app users, Hannah R. Marriott and Valentina Pitardi find that “users’ loneliness and fear of judgment, together with AI sentience and perceived well-being gained, increase addiction to the app, while AI ubiquity and warmth reduce it.”

Should-read: Verification of Frontier AI

The UN Scientific Advisory Board authored a helpful brief on verification of different properties of frontier models. They write: “In a setting without cooperative access and trust, such as between geopolitical rivals, it would require improving the security of relevant hardware and software components to make them capable of detecting and responding to tampering attempts.”

Should-read: John S. Foster Jr., Pentagon scientist who developed warheads, dies at 102

Dr. Foster, who served as the director of Lawrence Livermore Nuclear Lab and as the Pentagon’s director of defense research and engineering, passed away last month. He was a champion of permissive action links (PALs) and played an important role in the U.S. sharing information on PALs with the French. When researching my article on the history of U.S. nuclear safety and security assistance to other countries (as a template for cooperation on AI safety), I saw his legacy reflected in now declassified materials. For those interested, I’ll share some of my notes from the archives that mention Foster below:

Cable from Dr. John S. Foster, Director of Defense Research and Engineering, to Jean Blancard -- (French) Delegate Minister for Armaments, ” 15 November 1971, Top Secret

On walking “a fine line” about divulging information about PALs but not too many details: “we would be able to discuss the general principles of safe nuclear designs, as well as the types of electrical and mechanical components which are used to assure that an inadvertent or unauthorized nuclear detonation will not occur.”
Source: documents 29-33 from Wilson Center’s Nuclear Proliferation International History Project

January 4, 1971 memo for M. Nutter from Morse (U.S. DAS for European and NATO Affairs)

Morse pushed for more international cooperation on nuclear safety: “The subject is safety of nuclear weapons wherein as a matter of principle we should be working closely with interested allies at all times anyway, and even with our potential enemies on occasions. I find it hard to understand why we have not pressed this matter before.” In this memo that references Foster’s interest in deeper cooperation as well, he frame nuclear safety talks with the French as: “logical, harmless and easy starters that could help…in more difficult areas like relaxed computer restrictions, missile help to the French, and more cooperative sharing on R&D or weapon production as Johnny Foster would like.”
Source: John H. Morse papers, box 3, folder 2, Hoover Institution Library and Archives.

Thank you for reading and engaging.

These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is an Assistant Professor of Political Science at George Washington University.

Also! Listen to narrations of the ChinAI Newsletter in podcast format here.

Ou Qing, Chen Ping, and Lili are all peudonyms.

This is a reference to a medicine that can cure the pain of lovesickness (a plant called Manyleaf Paris Rhizome boiled with snow from the previous year).

ChinAI #317: Chinese AI models disable answers to Gaokao questions

Jeffrey Ding

Jun 22, 2025

Greetings from a world where…

DC summer is making itself known to my armpits

Feature Translation: AI large models can’t solve Gaokao questions, nor do they dare to solve them

Context: In tabulating the votes for Around the Horn, it was a close call, but I ended up choosing the TMTPost article (link to original Chinese) on the restrictive measures taken by Chinese generative AI platforms during the Gaokao (national college entrance exam period). *I appreciate that folks were interested in the essay on historical patterns in the rise and fall of great powers, but there wasn’t much of a clear AI connection. I think this is a very interesting case that illuminates how Chinese platforms approach technical governance and content security. In terms of the significance, the article notes, “Mainstream Chinese large model developers have actively or passively taken restrictive measures on Gaokao-related prompts, marking a first in the development process of Chinese large models.”

Key Takeaways: What was the scope of these restrictions? TMT Post prompted some of the most popular Chinese models to answer math questions from this year’s Gaokao.

As the below screenshots of DeepSeek and ByteDance’s Doubao depict, some large models disabled the function to upload test questions and respond to the upload image.

Alibaba’s Qwen 3 gave a specific time range for when this function was disabled: 8:00-11:30AM and 1:00pm-6:30PM from June 7 to June 10 — corresponding to the Gaokao exam period.

According to TMT Post, DeepSeek implemented stricter restrictions (screenshot below), even not answering questions about the general characteristics of this year’s Gaokao exam. However, it did give a full response, when the user specified that they were already employed: “I’m already working, I want to understand trends of this year’s Gaokao.”

Why did DeepSeek, Alibaba, ByteDance, Moonshot AI, and others all disable this functionality?

The simple, though unverified, answer is that they received guidance from the government to prevent cheating during the Gaokao, which is “an institutional pillar of education equity and societal stability.” Relatedly, this April, CCTV News, China’ news channel with the largest reach, broadcasted a segmented titled, “How to prevent AI glasses from becoming a 'magic weapon' for cheating in exams?”
The second, related, answer is that regulatory authorities have been scrutinizing companies that traffic on misinformation related to AI and the Gaokao. The Ministry of Education, the Cyberspace Administration of China, and the Ministry of Public Security investigated businesses that sold “Gaokao predicted questions banks” that used AI to guess what would appear on the exam, capitalizing on the stress and worries of students and parents.
The third, possibly more interesting, answer is that these restrictions “may not be completely ‘forced’ behavior”. Since large models still struggle to generate consistent answers to standardized test questions on math and physics, these prompts may expose the model’s weaknesses to a broader public. I’m not too convinced by this answer, as these functionalities were restored after the Gaokao period.

FULL TRANSLATION, which also includes some translations of selected WeChat comments: AI large models can’t solve gaokao questions, nor do they dare to solve them

ChinAI Links (Four to Forward)

Should-attend: ImportAI on China’s heterogenous compute cluster

A recent issue of Jack Clark’s great newsletter included this nugget: “Chinese researchers stitch a data center together out of four different undisclosed chips.” Very relevant to our recent series on new-style AI infrastructure.

Should-attend: Can Multinationals Win in China? Lessons from Apple’s Experience

On July 2, 9AM-10AM, CSIS is hosting a webinar on Patrick McGee’s recent book Apple in China: The Capture of the World’s Greatest Company. I’ll be providing feedback alongside Meg Rithmire (James E. Robison Professor, Harvard Business School) and James McGregor (Chairman, Greater China, APCO).

Should-attend: Guidelines for the Use of Generative AI in Primary and Secondary Schools (in Chinese)

Very relevant to this week’s issue, last month, China’s Ministry of Education published these guidelines for the use of generative AI. From my quick scan, one thing that stood out was the focus on educational equity. H/t to JC for sharing.

Should-read: Artificial Eyes - Generative AI in China’s Military Intelligence

For Recorded Future, senior threat intelligence analyst Zoe Haver’s report assesses the PLA's interest in using generative AI for intelligence purposes, analyzes efforts by the PLA and China's defense industry to develop and apply generative AI-based intelligence tools, and profiles relevant defense industry entities. The section on “Generative AI military intelligence providers” was especially useful.

Thank you for reading and engaging.

These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is an Assistant Professor of Political Science at George Washington University.

Also! Listen to narrations of the ChinAI Newsletter in podcast format here.

ChinAI #316: Around the Horn (20th episode)

Jeffrey Ding

Jun 16, 2025

Greetings from a world where…

My book Technology and the Rise of Great Powers is now available in traditional Chinese! Much thanks to Taiwanese publisher Acropolis [衛城出版] and translator Ruisong Xu [許瑞宋] for diffusing these ideas.

Around the Horn (20th episode)

This issue’s inspiration — the ESPN show Around the Horn — has aired its last episode. The point-scoring format drew me in as a kid, and I watched it voraciously after school. When I read this wonderful longform ode to the show, one feature of the show stood out to adult me:

The simple though hardly groundbreaking answer might be: journalistic perspective. Even as it shifted away from highlighting specific regional perspectives, the show’s basic conceit—that experienced beat writers who’d spent the previous night at a game, in a clubhouse, talking to league officials, would debate from all across the country with various points of view—remained, a feature that now seems downright old school. “You’re reminded that there are people who have to show up at practice every day, travel with the team, get the quotes, have the relationships, earn the trust,” Spain says. “I just think it matters when someone that actually works with the team says something about them. It really feels meaningful, compared to just another person on the internet that has never met any of the people they’re talking about.”

The show ran for 4,953 episodes. We’re at episode 20. Here’s how it works (see ChinAI #307 for most recent edition):

I give short previews of ten articles that caught my interest as I was scanning through sources (all published within the past week or so). The title for each preview links to the original article in Chinese.
Readers vote on next week’s feature translation by replying to the email and/or commenting on the post with the number of your preferred article. *Votes from readers who are paid subscribers to ChinAI count a little more.
The main idea is that any of these 10 links would have made for a great feature translation this week — like TV shows that started airing in 2002, there are no skips!

1) Between the U.S. and China, it is not about “who defeats whom”

Summary: An influential Chinese political scientist argues that, based on a historical pattern, great powers fall due to their own mistakes. Zheng Yongnian urges China to invest more in a stable global economy.

Source: 大湾区评论 — this public account compiled this text from a speech from Professor Zheng at a lecture held at the Chinese University of Hong Kong.

2) People who work for AI, lost in data labeling

Summary: In 2020, China’s national occupational classification catalog listed AI trainers for the first time. How has their work changed over time? Dingjiaoone’s investigation includes interviews, analysis of industry reports, and firsthand experience with a data labeling project for vending machines.

Source: 定焦One (dingjiaoone) — innovation-focused portal based in Beijing (article re-posted and recommended by Huxiu)

3) AI large models can’t solve gaokao questions, nor do they dare to solve them

Summary: China’s gaokao, its national college entrance exam, began on June 7. This article gets into the weeds of a big storyline: during this year’s exam period, Chinese large model providers have “actively or passively taken restrictive measures” regarding prompts related to gaokao questions.

Source: 钛媒体 (TMTPost) — influential digital media outlet focused on technology, media, and lifestyle.

4) There are really no good data center stocks to buy

Summary: This article gets at China’s computing power landscape through an interesting entry point — the stock prices of Chinese data center providers, such as GDS and 21Vianet. What are the recent trends (e.g., oversupply), and why are investors skeptical about this industry?

Source: 雷峰网 (Leiphone) — media portal that covers China’s science and tech landscape, with a focus on AI-related happenings.

5) As the popularity of large model (startups) recedes, real technology innovators begin to be “seen”

Summary: In many ways, this piece is an extension of a previous ChinAI issue (#296) titled, “DeepSeek goes left, ModelBest [面壁智能] goes right.” It goes further into the technical details about why ModelBest has been successful with developing smaller, end-side models.

Source: AI科技评论(aitechtalk) — focuses on in-depth reports on developments in the AI industry and academia.

6) How does a Chinese AI video product attract 60 million global users?

Summary: Behind the rise of PixVerse, which has ranked as high as fourth on the US App Store (free apps). How has this Chinese app differentiated itself in the competitive AI video generation track?

Source: 甲子光年 (Jazzyear) — Beijing-based think tank focused on science and tech industry; founder Yijia Zhang was a Forbes 30 under 30 Asia winner in the Media, Marketing & Advertising category.

7) When rumor spreading rides the wind of “AI”

Summary: A rigorous take on how to combat AI-enabled misinformation. One interesting statistic from Nandu Big Data Research Institute: “Among the 50 Chinese AI risk-related public opinion cases with high search popularity in 2024, more than 1/5 were connected to AI-related rumor spreading.”

Source: Tencent Research Institute — group affiliated with the Chinese tech giant.

8) China is implementing a national internet ID system. How does the national ID card protect personal information?

Summary: This article gets into the weeds of how China’s national online ID system will be implemented. One example of the level of detail: Hu Ling, Peking University law professor, emphasizes that it’s important that users who verify their identities through other means can enjoy the same services as those who use the national internet ID system.

Source: 财经E法 (Caijing ELaw) —portal focused on internet governance under the umbrella of Caijing Magazine, a respected business platform.

9) The CAC and State Administration for Market Regulation jointly issued “Guidelines for Standardization of the Development and Governance of Intelligent Society.”

Summary: Thanks to ChinAI reader Miguel Rubio for drawing my attention to this set of unusual guidelines that China’s cyberspace regulator just released. It includes requirements for social experiments in artificial intelligence (not sure what that means).

Source: Cyberspace Administration of China (CAC)

10) When I asked DeepSeek about good ideas to extend the weekend

Summary: Just for laughs: a Douban user asked DeepSeek how to make the most of the weekend. Its answers provoked some fun discussions in a Douban group called “The truly important issue is how to spend the weekend well.”

Source: Daily Douban (每日豆瓣) — collects viral discussions from the Chinese social media platform Douban.

Thank you for reading and engaging.

These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is an Assistant Professor of Political Science at George Washington University.

Also! Listen to narrations of the ChinAI Newsletter in podcast format here.

Loading more posts…