,
15 tweets,
3 min read
Read on Twitter
I've spent the past month writing a thing on AI in China, some takeaways:
just like anywhere else, AI 人工智能 means everything & nothing in China, spanning core computational technologies and theoretical research as well as mundane applications of algorithms and interface gimmicks - but at least the Chinese term for it includes 人 (man) and 工 (labor)
the AI hype kicked off in China in 2016, and it is connected to Google's blow to ethnonational board game pride - fallout from AlphaGo's victory set in motion a whole chain of policy plans, venture capital investment and popular media enthusiasm
in spite of self-congratulatory narratives emerging around new policy documents, the Chinese government wasn't particularly foresighted about the technology, as it basically jumped on the industry bandwagon once it was clear that AI was going to impact society & economy
speaking of plans, AI policy and industry hype both follow quite closely the regular pattern of tech innovations in China, re-enacting previous cycles (Web 2.0 in 2009, mobile apps in 2012, Big Data in 2014, etc.) and latching onto longer history of national tech planning
this is clearly visible in the fractal nature of most policy goals - as the year of "complete informatization" draws near (2020), AI plans move the goalpost to 2025 (achieving major breakthroughs) and 2030 (achieving world leadership)
on the ground, AI is chiefly deployed in a couple of ways: content curation and business optimization - on the interface side, to lock in users through personalized feeds and recommendations, and on the cloud side, to compete with the rising cost of labor and slowdown pressure
speech recognition/synthesis and computer vision are described as the fastest growing sectors of the AI market, and this can be traced to two things: a) the labor pool for data labeling, and b) the military & security demands for surveillance systems
other sectors remain in the catch-up stage and are mostly showcased through tech demos and promotional fronts (autonomous vehicles, smart retail), but the actual breakthroughs (if any) will come from hardware, as local industry is pushed to develop homegrown AI chips
Chinese authorities and companies are complicit in both modeling national AI growth on the U.S. and imagining themselves as locked into a bipolar arms race with their competitor, while the reality of R&D is complicated by global cooperation and international coordination
the geograpy of the Chinese AI industry is indicative of its economic implications: BAT and largest players based in Beijing/Shanghai/Shenzhen, research centers often co-located in China and abroad to maximize talent acquisition, data labor hidden in low-tier cities & countryside
just as in previous tech hype cycles, local governments are happily following central signaling and dumping money into innovation districts, startup incubators and AI funds, with each city and province claiming a specific segment of the industry
the elephant in the room is, as expected, the Social Credit System, which is both an ideal tool of automated governance and a real Stack of financial rating entities - it is deeply connected to AI and digital platforms, but very few reports mention it
China's most recurring claim to AI leadership is grounded on data primacy - China being the world's largest data producer, and its billion of users functioning as inexaustible data faucets - but there is little evidence that data alone will support AI growth in the long run
among the 50+ documents I've read, @jjding99's report, articles & newsletters are the most informative, so follow them
