ChinAI #376: Around the Horn (27th episode)
Greetings from a world where…
Overview
Greetings from a world where…
Slow Horses is back
…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).
Around the Horn (27th episode)
At least we still have PTI. Let’s honor Around the Horn again.
This is our 27th roundup of articles on China’s AI ecosystem. For new readers, here’s how it works (see ChinAI #365 for the previous edition):
I give short previews of ten articles that caught my eye during a scan through my usual 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. *Extra weight goes to votes from those of you who support ChinAI through a paid subscription.
The main idea is that any of these 10 links would have made for a great feature translation this week — like the Hadestown soundtrack, there are no skips!
1) My “Tier-5” hometown became a Token Capital
Summary: Both Goldman Sachs and Wired spotlighted Ulanqab, a city in Inner Mongolia, as a new center of China’s AI computing boom. This reporter visits their hometown to uncover what’s actually happening beneath the headline figures.
Source: 量子位 (QbitAI) — news portal that regularly covers AI issues, similar to Leiphone. They also publish longer reports.
2) Using AI for National Day travel: the first wave of users feel let down
Summary: During this one-week holiday for National Day, Chinese travelers offshored more of their planning to AI travel assistants. Disappointments ensue.
Source: 虎嗅 (Huxiu), a well-known platform that shares user-generated content but also publishes their own pieces on China’s science and technology ecosystem.
3) Why do Kimi and DeepSeek appear in the same model configuration?
Summary: This one would push the bounds of my technical knowledge. The article breaks down a trend in Chinese large model architectures: for some layers, they follow Kimi’s linear attention approach; for other layers, they mix in DeepSeek’s sparse attention approach.
Source: AI科技评论(aitechtalk) — focuses on in-depth reports on developments in the AI industry and academia.
4) BAT giants redo AI pre-training to fix the “Dirty Data” pitfall
Summary: Why have so many Chinese AI labs reworked their pre-training process in the past six months? Dirty data. This article explores junk portions of corpuses, ambiguous data labeling rules, and evaluation set problems.
Source: 雷峰网 (Leiphone) — media portal that covers China’s science and tech landscape, with a focus on AI-related happenings.
5) Alibaba DAMO Academy’s Medical AI, piecing together a decades-long picture
Summary: We’ve covered the frustrations and challenges of medical AI applications in China in previous issues. This pieces profiles the long road to success for Alibaba’s medical AI team.
Source: 晚点 (LatePost) — publishes in-depth analysis on China’s tech scene, known for deeply-sourced reports.
6) Jev Dethroned: Chinese open-source decision model StartLux takes top spot
Summary: It was fun to learn a little more about decision models like TypeSafe AI’s Jev, which return judgments about different choices/scenarios, instead of open-ended text. How does the Chinese open-source model StartLux compare?
Details
Source: 机器之心 (jiqizhixin) — media portal similar to Leiphone in coverage.
7) Research Report on Development of the Token Economy (2026)
Summary: It’s been a while since we’ve dug into a CAICT white paper. This one on the token economy looked informative.
Source: 中国信通院 (CAICT) — a think tank under China’s Ministry of Industry and Information Technology.
8) In the US-China AI race, the deciding factor isn’t compute, it’s another thing
Summary: This essay argues that U.S. hegemony in computing power is built not just on chips; rather, the key factor is the ability of Google, Microsoft, and other hyperscalers to build large-scale infrastructure that aggregate computing power, land, energy, and cooling resources (including in other countries like the Netherlands)
Source: 文化纵横 (Beijing Cultural Review) — leading platform for contemporary political and cultural thought, which also publishes a quarterly journal.
9) What happens when AI employees join the workforce?
Summary: Tencent convened a roundtable on AI employees, with a focus on what the transition from pilots to large-scale deployment. The discussion draws on insights from Tencent’s WorkBuddy agent.
Source: 腾讯研究院 (Tencent Research Institute) — Tencent’s policy and governance research arm.
10) Ban on AI-Human “Romantic Relationships”: New rules on minors’ internet use to tighten controls on virtual companions
Summary: Building on companion AI regulations from a few months ago, new draft regulations on safeguarding minors’ internet usage impose further restrictions on AI services. Article also cites previous investigations by Southern Metropolis Daily on AI apps’ weak child protection measures.
Source: 隐私护卫队 (Southern Metropolis Daily’s Privacy Guard Team) - connected to Nandu Personal Information Protection Research Center.
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).
Any suggestions or feedback? Let me know at chinainewsletter@gmail.com or on Twitter at @jjding99
Source
Originally published at chinai.substack.com.
