Read on Twitter
Was initially excited to see this piece on China's AI & tech surveillance from @anderson @TheAtlantic. But it too glibly repeats a lot of China's own hype about surveillance systems w/o enough tough scrutiny. And it needed better fact-checking. Thread:
https://www.theatlantic.com/magazine/archive/2020/09/china-ai-surveillance/614197/
1/n
https://www.theatlantic.com/magazine/archive/2020/09/china-ai-surveillance/614197/
1/n
First, the idea that the surveillance state was born + built in Xinjiang, & is now being exported, is wrong. These systems were pioneered in the urban east. Xinjiang was *behind* the curve -- and then leapfrogged. Very different story than presented here.
2/n
2/n
Second, presenting China as unbroken arc of info + control is weird, esp Chiang Kai-shek. My first book talked about how Chiang *didn't* effectively use baojia, etc to manage the threat of communist rebellion--which is why he lost control of mainland China?
3/n
3/n
Third example: China is not just now turning to Latin America to export surveillance tech! They were among the first countries to adopt Safe City type platforms--before Africa, not behind it.
4/n
4/n
It would be easier to overlook basic errors if there wasn't massive reporting & research easily accessible in open sources for fact-checkers to look at.
See for example, this @BrookingsFP report:
5/n https://www.brookings.edu/research/dealing-with-demand-for-chinas-global-surveillance-exports/
See for example, this @BrookingsFP report:
5/n https://www.brookings.edu/research/dealing-with-demand-for-chinas-global-surveillance-exports/
Let's also clarify: not all surveillance technology is AI-enabled. Frankly, some of the whiz-bang tools featured here aren't even connected to basic data integration platforms -- let alone sophisticated AI. We need a national conversation that can keep this straight.
6/n
6/n
Why does this matter? Because these systems are aimed at deterring certain kinds of behavior. To do that, they convince ppl they are infallible/all-powerful. Can we *please* not do the CCP's deterrent work for them by overstating capability? 7/n
What are the issues?
- Massive bureaucratic fragmentation.
- Power struggles over who wins/loses in design of inter-agency data integration.
- Police forces not accustomed to policing-by-app.
- Lack of human capital to analyze data being collected.
8/n
- Massive bureaucratic fragmentation.
- Power struggles over who wins/loses in design of inter-agency data integration.
- Police forces not accustomed to policing-by-app.
- Lack of human capital to analyze data being collected.
8/n
To be effective, we need to talk not just about these systems' successes, but gaps, problems, shortcomings. Algorithms can be wrong or biased. They can have false positives. They can be logical but unveridical. Those have major political consequences too.
9/n
9/n
That was supposed to be by @andersen @TheAtlantic. Apologies for the mis-tag.
