11-week intensive Chinese program (ICLP) to move from B2 to C1 Mandarin and remove the bottleneck on US-China-Taiwan AI governance research I'm already doing
Endorsed by+1
11-week intensive Chinese program (ICLP) to move from B2 to C1 Mandarin and remove the bottleneck on US-China-Taiwan AI governance research I'm already doing
Endorsed by+1
Project Details
Updated 07/14/26 · Provided via application · Verified
What I'll do: I'm requesting funding for NTU's International Chinese Language Program (ICLP) for the Fall quarter (Sep 23 - Dec 11, 2026). ICLP is Taiwan's most intensive Mandarin program and historically the training pipeline for US diplomats and China analysts. It includes 4h of classes and ~6h of homework daily, and my goal is to upskill from my current B2 to a C1. This would enable me to engage fluently with Chinese-language AI governance sources for my ongoing work, as well as to work across Taiwanese and Chinese institutions, and translate between the international AI safety community and local actors. Attending ICLP is the single most impactful thing I can do to next get there.
Who's involved: solo project.
Concrete output: passing HSK 6 and TOCFL 5 by January 2027 is the criterion for success: it would demonstrate a C1 level in both Traditional and Simplified Chinese, and would allow me to fluently engage with Chinese-language AI governance sources from both Taiwan and China.
Theory of Impact
Updated 07/14/26 · By grantmaking.ai
Any US-China agreement or verification regime over frontier AI depends on granular knowledge of who shapes Chinese AI regulation, and on people reading that regulation in the original. The February 2026 Summit on Existential Security survey of AI safety leaders found that China-US relations was one of the sub-fields "most needing significantly more resources." Researchers embedded in Taiwan (chokepoint of any compute-based verification scheme) are close to nonexistent in the AI safety ecosystem. This grant closes the gap between my current position and that profile, upskilling my Chinese from B2 to C1 in one quarter.
China pathway: I'm currently a GovAI Summer Fellow, supervised by Jeffrey Ding, mapping the authorship of Chinese AI regulation explainers for an audience of Track 1.5/2 dialogue conveners - a project initially proposed by Matt Sheehan. Researchers convening Track 2 dialogues have told me that this would be directly useful to their convening work.
Taiwan pathway: how Taiwan manages chip production shapes the conditions under which advanced AI develops. I've held preliminary conversations about chip verification mechanisms for Taiwan, following ~6 months of research and field-building in Taiwan: interviewing Taiwanese AI governance institutions, on my Substack, the NTU AI safety group, giving feedback to Taiwanese applicants of AI safety fellowships, and liaising with TARA and ML4Good to improve their reach in Taiwan.
Jeffery Ding, Matt Sheehan, and Scott Singer are three of the leading scholars in China-AI governance space. Jeffery have had testimonies in Congress.
The project that David is working on will be one of the capstone for track 1.5 and track 2 dialogue to discuss about. If done well, it can highly improve the possibility of two sides cooperation on achieving meaningful outcome. US-China cooperation on AI governance solidifies the AI governance globally.
Sponsoring David for his language learning project is sponsoring advancing the knowledge on China-AI development in the english world and furthering the possibility of US-China cooperation.
The constraint: at my current B2 level (TOCFL advanced-high reading, 585/700), I can do the above work slowly and with friction. With a C1, I could read policy documents in Chinese at working speed and communicate in Mandarin fluently.
Why ICLP: ICLP is the only program in Taiwan that can take me from B2 to C1 in one quarter, with its 4h of daily classes in groups of 1-4, plus ~6h of homework.
Verifiable: the grant has fixed cost and fixed dates (Sep 23 - Dec 11), and success can be verified by two standardized exams in January (HSK 6 and TOCFL 5).
Costs: given ICLP's intensity, I won't produce research during this quarter, but I will continue my co-facilitation at the NTU AI Safety group (which doubles as Mandarin practice), and contribute to TARA's round 2 through talks to participants.
Scott Singer (Oxford China Policy Lab, Carnegie) recommended ICLP to me directly, and he studied there. Jeffrey Ding, my GovAI supervisor, also endorses the program.
From January 2027, I'll apply to full-time AI governance roles with the profile that the gap above called for: C1 Chinese, the GovAI fellowship completed, and the dataset published.
Counterfactual: without this funding, I'd continue my AI governance research during fall, with friction, and alongside part-time teaching and facilitation work. The language gap would close through self-study over years rather than in one quarter.
Jeffery Ding, Matt Sheehan, and Scott Singer are three of the leading scholars in China-AI governance space. Jeffery have had testimonies in Congress.
The project that David is working on will be one of the capstone for track 1.5 and track 2 dialogue to discuss about. If done well, it can highly improve the possibility of two sides cooperation on achieving meaningful outcome. US-China cooperation on AI governance solidifies the AI governance globally.
Sponsoring David for his language learning project is sponsoring advancing the knowledge on China-AI development in the english world and furthering the possibility of US-China cooperation.
Best wishes