Fake it til you make it. Journalist, Polcom PhD at @stanford. Also @stanfordhai @KnightHennessy, formerly @latimes @PKU1898 @AmherstCollege

Palo Alto, CA
Joined July 2010
Matt DeButts retweeted
A new piece from me on @NYMag is out today. It is the second installment in a reporting trilogy I’m calling “AI and the Chinese society.” Earlier this year, I visited China’s largest get-rich-with-AI company, which sells AI courses for profits. It has more than 80,000 paying members. Its meeting rooms have names like “Apostle of the Method” and “Evangelist.” Their membership system resembles a pyramid scheme. I was greeted by Mr. Tree, an in-house influencer and AI educator whose social media bio promises to “search for 100 business answers in the age of AI”—an echo of Beijing’s “AI Plus” plan. The company felt like a prosperity-gospel church, and salvation meant adopting AI as fast as you can. In Liangzhu, China’s vibe-coding capital, which I wrote about, young people give one another AI lessons for free in their living rooms. I also met Xiaokai, a 25-year-old filmmaker who became a pro on AI video generation through free classes, built a business earning several million yuan in profit, and now walks red carpets in Shanghai and Beijing. To understand this AI education craze, I found peoples’ fear of being left behind, their religious faith in education, and millions of ordinary people trying to find a place for themselves in a AI future Beijing is summoning on behalf of everyone. ----- Earlier this week, @KaiserKuo and I chatted about this trilogy on the great @SinicaPodcast in detail and some behind-the-scenes stories. You can listen to it now (links in the comments)! Also. Many thanks to my wonderful editor, Joy Shan; my fact-checker, Jessica Suriano; and some friends who read the draft and offered feedback.
7
69
9
367
27,609
Matt DeButts retweeted
My first piece for @NewYorker is out today! As Xi Jinping lands in DC to start his state visit, you'll hear a lot about the US-China AI race. I'd like to tell a complementary AI story. This past May, a few days after Trump visited Beijing, I went to the town people call China's vibe-coding capital, Liangzhu. There, you learn that China's AI scene is not monolithic. It's also full of contrarians, cool vibe coders, spiritual seekers, and people who love AI but refuse to join the 996 grind. Liangzhu is on the edge of Hangzhou. I hung out with a group of anti-mainstream builders who use AI to make whimsical apps for fortune-telling, mental well-being, and creativity. Some of them run their lives entirely by bazi astrology (their founder almost became a Taoist monk). They meet on "Crazy Thursday" to vibe-code together. Rent is cheap, so they can get a bigger place with a living room to have friends over (unthinkable in Beijing or Shenzhen). In Liangzhu, they want to use AI to forge their own future. It’s a ground-level story—less about model capabilities, hyperscalers, or AI policy, and more about how people in China are using AI, and the human stories behind it.
49
195
36
1,175
184,530
Matt DeButts retweeted
When the state cannot afford to jail activists, it calls their parents. My research theorizes Emotional Proxy Repression: authoritarian states exploit the emotions bound up in intimate relationships, obligation, care, and shame, to pressure political actors with full deniability. Patriarchal family norms make the feminist activists I study especially vulnerable to this form of repression. I see my work as a political intervention, a refusal to let privatized state violence stay hidden. Open access: doi.org/10.1016/j.wsif.2026.…
7
23
1,914
Such an interesting idea. Leave it to @afrazhaowang to turn a coffee chat into a published article in a few weeks. So impressive, as always!
My piece on AI and creative writing, A New Kind Of Creative Poverty, is out in @NoemaMag, edited by the wonderful @latams The relationship between creative writing and AI is a well-worn subject by now, but i couldn't stop thinking...why can't great writing be measured, quantified, or converted into the empirical alignment problems of post-training? Why is "greatness" so illegible, even about itself? The question sent me back to my first (accidental) encounter with the French modernist Charles Baudelaire in China as a kid. In this piece, I also write about the increasingly stifling Chinese internet, the place I grew up, what people in Taiwan derisively call the “Simplified Chinese Internet” (简中互联网). I see a resemblance to LLM post-training, both operate through reward and punishment, and both breed homogeneity. The essay is indebted to @zhengdongwang's blog; to a conversation with @danwwang in Shanghai about the Simplified Chinese Internet; to @jasminewsun's writing on why LLMs make bad writers but good editors; to a talk with @debutts on reward-and-punishment in LLM training and the Chinese internet; and to my accidental part in @chrisbest and Oliver Habryka's exchange on the meaning of human writing at the Manifest conference (which @hsu_steve invited me to speak). A stretch of reading on model collapse fed it too, as well as a discussion with @realJerryZhou over korean food. the draft was read and improved by @mynameisceline and several other wonderful researchers and writers. link in comment
1
231
Can't wait to read!
Woo first copies are here! Our Hearts Burned For Home, on sale very soon October 12. The book unearths the untold story of Uyghur exiles who fled China’s expanding police state—only to become a pivotal force behind the fall of Bashar al-Assad’s regime. globalreports.columbia.edu/b…
1
316
Today I start a new job as a White House correspondent with NPR. I am so excited; this is one of the biggest beats in the world right now, and a challenging one at that. npr.org/people/705255512/emi…
78
38
6
765
63,935
(1) New research with @ttzhang107 in @chinaquarterly: China’s digital complaint institutions (kind of like a “fix-this-problem” hotline) are being heavily used — and are increasingly effective. In 13 years, govt resolved almost 2x as many citizen complaints, from 17 to 31% (1/6)
1
4
6
238
(5) We used double-coded human classification with DeBERTa to assess govt resolution, and anonymized IP address look-ups + geospatial tools to identify growth. For more info, check out our paper here cup.org/4z320w3 and follow my co-author @ttzhang107 (5/6)
1
1
36
(6) And lastly thanks to Ji Feng and Melody Wang, as well as @StanfordSCCEI @StanfordHAI and @AU_SPA for supporting our work! (6/6)
39
It's a crying shame that there aren't more intl. correspondents accredited in China today. All the best to @dakekang and let's hope AP will get more visa approved soon.
You’re unable to view this Post because this account owner limits who can view their Posts. Learn more
1
2
25
3,974
Absolutely not surprising but a limited leak of a Chinese local surveillance platform drawing off an apparently large national one has a special drop down for tracking for tracking foreign journalists in China… including me. @NetAskari found the data open.substack.com/pub/netask…
34
148
10
628
57,791
Matt DeButts retweeted
Using Chinese AI models for coding be like:
NEW: House committees are probing Airbnb and Cursor parent Anysphere over their use of Chinese AI models @semafor
23
716
134
11,172
1,246,507
Matt DeButts retweeted
If you are interested in academic research on this very subject, check out @ucdpolitics PhD candidate @LinetteMLim here: journals.sagepub.com/doi/ful… and here: tandfonline.com/doi/full/10.…
You may not like to hear this, but there is a correlation between improving views of China globally and Beijing’s expulsion and harassment of foreign journalists. The Chinese Communist Party has successfully shaped the portrayal of China globally, by severely restricting journalists ability to analyze and report on the country. The result? Higher favoribility ratings of Xi Jinping and China than Trump and the United States, and growing favoribility rates of China in the United States. This is not to downplay the clear, Trumpian reasons the U.S. has lost global luster — rather, it’s to express consternation and concern of entities around the world for believing China to be a responsible global actor. fccchina.org/2026/04/20/fccc…
9
16
3,943
Striking statement from China’s foreign press club, explicitly making clear its critique of China revoking visas, China detaining journos over coverage, not associated w/any media outlet. Media orgs - powerful ones - are not speaking up, and they should. fccchina.org/2026/04/20/fccc…
47
85
7
218
46,001
Excited to share that the Mandarin-language version of my book is being published April 1 in Taiwan !
5
15
2
101
8,155
Matt DeButts retweeted
I’m hiring a predoc starting this summer. The focus is building AI workflows to improve research efficiency. More importantly, I value independent thinking and ambition. You do not need to fit into a narrow topic. If you’re interested in applied statistics, causal inference, or Chinese politics, broadly defined, plz apply. You can propose your own project --- I will support it. iriss.stanford.edu/predoc/20…
9
45
2
167
21,960
Matt DeButts retweeted
The quirky Ma Qianzu explains why the Chinese internet is so frustrating to read: there’s little actual analysis, only a straightjacketed state media and a proliferation of dramatic narratives And a metaphor from a great interview by @afrazhaowang afraw.substack.com/p/first-d…
7
13
133
14,606