From the Concurrent substack, September 11, 2025:
On circles, misconceptions, deep infrastructure, founder archetypes, state's role, memes, AGI. A conversation about the topology of China’s AI world for you.
H is an AI engineer at one of China's largest tech companies, leading an open-source project. A few weeks ago, I met him in Shenzhen during a self-organized "China AI/tech tour" with several US tech writers.1 After peppering him with questions—the most obvious one being "What's Silicon Valley's biggest misunderstanding about China's AI?"—I discovered someone so thoughtful and genuinely bilingual.
His daily routine reveals the paradox: wake up, doomscroll X, debate on LessWrong, drive his NIO to work, then use English more heavily than when he studied in California. He loves watching Westworld and treats Gödel, Escher, Bach as AI scripture. Apart from geography, H could be any engineer at Anthropic or Google. He represents the global open-source AI community: following identical cultural patterns, reading the same viral posts, grappling with the same technical problems, laughing at the same memes. He just happens to work in China.
"'Chinese AI' is merely a geopolitical label," H said. "All LLMs fundamentally compress and regenerate collected data." Yet despite this universalist perspective, he articulately mapped the profound topological differences between Chinese and American AI ecosystems while understanding the China model's essence with surgical precision....
....MUCH MORE
