ChinAI #285: Chinese LLMs go global
Greetings from a world where…
my Hawkeyes have a chance to sneak into the college football playoff
…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: Coming back from “internal injuries,” AI applications are opening up overseas
***Thanks to Marianne Lu for contributing this week’s feature translation and analysis. Marianne is a technology and security policy fellow at RAND. She recently graduated from Stanford University, where her thesis focused on U.S.-China AI competition. What follows is her analysis (lightly edited by me).
Key Takeaways: Chinese LLMs have done surprisingly well overseas.
At least three of the most downloaded AI apps in the U.S. are from Chinese companies.
This reflects a growing awareness among Chinese AI companies that “simply competing within the domestic market isn’t enough –– they must go global…In 2024, AI and international expansion have become two of the few remaining paths that still show growth.”
A variety of players are riding this wave of overseas expansion, including established tech giants, newer AI unicorns, and smaller startups.
International expansion offers a few key benefits, including access to broader sources of user traffic and better LLM technologies.
Entering new regions allows companies to bring in waves of users while avoiding the challenges of the domestic market, including competitors who imitate features, engage in price wars, and poach talent.
It also opens up access to superior technologies that are unavailable in China –– the overseas version of ByteDance's Coze, for example, supports GPT-4o.
Additionally, the overseas market is more receptive to paid subscription models and boasts a better funding environment.
But the picture is far from perfect.
The article admits that overall, the success rate for AI products that go overseas is low: “Very few can achieve instant success and maintain stable operations.”
One possible explanation is that AI development has not yet entered an era of “truly killer” applications: “Cautious observers believe that AI has not yet entered the ‘mobile internet era.’ The transformative capabilities of AI large models resemble the early days of internet connectivity –– we are still in the stage of improving infrastructure…”
Achieving scale, navigating commercialization, and ensuring compliance are some of the key challenges facing Chinese companies trying to go abroad.
Many companies struggle to get off the ground because of an inadequate understanding of overseas user needs.
Achieving profitability also remains a significant hurdle. While earlier developers earned their “first pot of gold” through minor innovations and AI shell applications, the market has become much more competitive.
Compliance is another issue. Some payment services providers have apparently broken off partnerships with Chinese AI products due to risk concerns, and new regulations such as the EU AI Act place stricter standards on Chinese AI companies venturing abroad.
The article concludes: “For Chinese companies going abroad, becoming the next TikTok still requires overcoming many hurdles. “
Analysis: Observers often highlight how Chinese government support and direction facilitate the overseas expansion of Chinese tech companies. This article sheds light on the role of domestic market dynamics in companies’ decisions to go abroad, as well as the considerable challenges that they face along the way.
FULL TRANSLATION: Chinese LLMs Go Global
ChinAI Links (Four to Forward)
Should-read: CAC solicits public opinions labeling AI-generated synthetic content (in Chinese)
Sourced from the Emerging Technology Observatory’s Scout tool: “On September 14, the CAC announced the solicitation of public comments on a draft of measures for the identification of AI-generated content. Individuals have one month to provide comments. The draft for comments defines AI-generated content and requires network service providers to explicitly label AI-generated files as they are downloaded, copied, and exported. The draft also requires that providers standardize the dissemination of AI-generated content and remind users to declare whether their files include AI-generated content.”
Should-read: Soda Science - A Conversation with Susan Greenhalgh
For the Made in China Journal, Yangyang Cheng interviews Susan Greenhalgh about how Big Soda exported its ideas about nutritional science to China. Some fascinating insights in this interview, including:
The US soda industry’s secret weapon was an industry-funded scientific non-profit called International Life Sciences Institute (ILSI). Based in Washington, DC, but with branches around the world…ILSI’s China branch was run by a very capable former high-level official from the Ministry of Health. Because of her political position and policy clout, ILSI-China was physically located in the Ministry and became the centre of obesity work in the country during the key years of 1999 to 2015. This extraordinary setup gave Big Food a direct hand in the making of China’s science and policy on obesity.
Two book promo links:
Apologies but the book promo hustle continues:
Technology and the Rise of Great Powers book talk at Stanford University’s Center for International Security and Cooperation: Tuesday, October 15, 2024; 12:00 PM - 1:15 PM (Pacific).
I had a great conversation about the book with Eleanora Mattiacci, Associate Professor of Political Science at Amherst College, on the New Books Network podcast. It’s a platform to discover new academic books!
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