Greetings from a world where…
zongzi is the king of glutinous rice dishes
…***Can you help ChinAI reach 200 paid subscriptions as we near the 20k subscriber mark? Write it off as a business expense! 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 (15th edition)
We are long overdue for another trip around the horn (see ChinAI #256 for previous iteration). For new subscribers, here’s what you need to know:
I give brief snippets of ten articles that could be ChinAI feature translation (all published within the past week or so). The title for each preview links to the original article in Chinese.
Readers pick next week’s feature translation. Just reply to the email and/or comment on the post with the number of your preferred article. *Votes from readers who are paid subscribers to ChinAI count a little more.
Main premise is that any of these 10 links would have made for a great feature translation this week — like Lorde’s Melodrama, there are no skips!
1) What is the “mother tongue” of LLMs?
Summary: The best large language models can handle prompts in numerous languages, but what language do they dream in? This article summarizes discussions about this topic, centered on a paper published by researchers from Ecole Polytechnique Federale de Lausanne in Switzerland.
Source: 新智元 (xinzhiyuan) — media portal that covers China’s AI landscape
2) The first anonymous Chinese large model arena ranking is released
Summary: The SuperCLUE team has announced a leaderboard for Chinese large language models, similar to LMSYS Org’s Chatbot Arena, which relies on user voting and feedback (in contrast to automated evaluations of LLMs across thousands of test prompts). The first results compared 17 large models and collected 12,600+ votes.
Source: CLUE中文语言理解测评基准 (SuperCLUE) — organization that tests the capabilities of large language models from Chinese and international labs.
3) Counting down the life and death of a star autonomous driving company
Summary: HoloMatic [禾多科技], an autonomous driving start-up founded by ex-Baidu machine learning scientist Ni Kai, has come to a crossroads. Why has this ballyhooed startup faced such struggles, and what does this tell us about the broader autonomous driving ecosystem in China?
Source: 雷峰网 (Leiphone) — media portal that covers China’s science and tech landscape, with a focus on AI-related happenings.
4) Latest developments in the “Stanford AI team plagiarizes China’s open source model” case
Summary: On May 29, a Stanford University AI team released a Llama3V model, claiming that it found a very cost-effective way to train a state-of-the-art multimodal model, comparable to GPT-4V and Claude Opus. This effort plagiarized an open-source large model released by the Chinese start-up ModelBest — coincidentally, we actually covered this company in a previous post (ChinAI #199) that was picked by Around the Horn voters.
Source: OSChina — portal that covers China’s open source community
5) China plans to introduce data property rights system
Summary: This article outlines China’s proposed data property rights system, which aims to encourage better circulation of high-quality data. It contains some good details about the scale of data resources in China as well as how this property rights system interacts with the data security law, PIPL, and other regulations.
Source: 财经E法 (Caijing ELaw) —portal focused on internet governance under the umbrella of Caijing Magazine, a respected business platform.
6) After the (LLM) price war, Zhipu AI has come to the next stop — commercialization
Summary: As tech giants like Tencent, ByteDance, and Alibaba have lowered prices on their large model services, one big question looms: how can startups like Zhipu AI compete? This article gets into the weeds of how Zhipu AI is trying to stay competitive and commercialize its large models.
Source: AI科技评论(aitechtalk) — focuses on in-depth reports on developments in the AI industry and academia.
7) Beijing Cultural Review’s “AI: More than just a Technological Revolution” Issue
Summary: This influential Chinese bimonthly magazine offers high-quality commentary on a wide range of topics, including AI. They recently did a recap of their previous four issues, and the April one was dedicated to AI. Article topics included: the concentration of AI power in a few tech giants, the geopolitical logic of data center expansion, and how workers’ organizations can adapt to generative AI.
Source: 文化纵横 (Beijing Cultural Review) — A previous ChinAI issue (#167) translated their article about the re-skilling of migrant workers in the context of China’s digital transformation.
8) This AI product has over 100 million users, ranking first in monthly active users…and some people secretly use it to make money
Summary: The hottest AI product in China is not a chatbot but rather Baidu Wenku, a one-stop shop for content creation and acquisition. At least, that’s according to usage statistics by aicpb.com [AI 产品榜], a very cool portal that tracks 10,000+ AI products across 100 AI application subfields.
Source: 量子位 (QbitAI) — news portal that regularly covers AI issues, similar to Leiphone.
9) How can voice actors whose voices have been stolen by AI keep their jobs?
I used the ETO Scout tool to source this pick for Around the Horn.
Summary from Scout: This article illustrates the challenge (posed by the growing use of AI for voice generation) using anecdotes and interviews with several Chinese voice actors. The Chinese voice acting market is also hyper-competitive with hundreds of low-skilled actors competing for a handful of minor roles in the studios in major cities…Some unscrupulous companies have recorded voice actors without their full consent and then incorporated their voices into AI-powered products; junior voice actors seldom have the resources available to protect their rights…
Source: Hedgehog Commune (刺猬公社) — influential account that covers the content industry
10) National Data Resource Survey Report
Summary: This white paper provides details on China’s data usage rates, computing power capacity, and data circulation rates. There are also sections dedicated to tracking developments in public datasets.
Source: 国家工业信息安全发展研究中心 (China Industrial Control Systems Cyber Emergency Response Team) — state-backed institutions that also runs database on cybersecurity vulnerabilities.
ChinAI Links (Four to Forward)
I’ve thrown a lot of text at you already, so let’s keep it brief with the recommendations this week:
China Law Translate’s excellent thread on the NDRC’s new Social Credit Action Plan for 2024-2025
For MIT Tech Review, Josh Goldstein and Renée DiResta break down OpenAI’s report that disclosed misuse of their products for influence operations.
CSIS’s Pekingology podcast episode with Yeling Tan about her thoughts on China, WTO, industrial policy, and more.
My wife and I watched “Hit Man” this weekend. Maybe 50% of my desire to watch this film was getting to read Adam Nayman’s review afterward.
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.
Any suggestions or feedback? Let me know at chinainewsletter@gmail.com or on Twitter at @jjding99
4, 9
7, 8, 10