China already uses facial recognition to develop racial profiles and closely monitor the Aigur minority
Since last year, the measures implemented by the Chinese authorities to control the revolts of an independentist and / or Islamist character supported by various sectors of the Uighur ethnic minority (a people of Turkic culture and majority Muslim faith) have caused considerable controversy among international organizations linked to the defense of human rights.
But a few months ago this controversy revolved around the existence of ideological “reeducation camps” that would house up to one million prisoners, now the headlines are the vast system of facial recognition launched by the Chinese government, which according to the International media would have been using artificial intelligence to generate ‘racial profiles’.
Using machine learning , engineers manage to train AI systems to detect what kind of patterns (sometimes invisible to the human eye ) are present exclusively in a certain group of images , be they photos of cats taken from Facebook or, as in this case, photos of Uighurs taken from police files.
That is to say, that China would be dedicating resources to identify people by their ethnic peculiarities (for now they would be focusing solely on the Uyghur ethnic group) and not by their individual traits , as they have been doing up to now.
This is relatively simple, since the appearance of the Uyghurs, similar to that of the peoples of Central Asia, differs greatly from most ethnic groups in the country , such as the majority ethnic group (the Han) or their neighbors, the Tibetans.
According to the New York Times, this technology “marks the beginning of a new era of automated racism” and is already allowing its government to monitor the comings and goings of 11 million Uighurs , both inside and outside the borders of their home region, Sinkiang. In fact, 16 provincial and regional police departments (for example, in the coastal region of Fujian) would have already adopted this new technology.
It is not that China is worried about keeping this program secret : CloudWalk, one of the companies that supply this technology, states on its corporate website that one of the possible uses of it is to set alerts (directed to the police) if in a A certain number of Uighurs appear in a given period of time.
The only consolation that remains for the supervised is that the system is still far from perfect: it still often fails , and its success depends on factors such as the placement of the cameras or the level of illumination.
In the West, where the debate on algorithms and races revolved around the presence of unconscious biases in the facial recognition systems that could be discriminating against monories, it may seem incredible this clear and unnoticed bet of algorithms of racial segregation by of a government.
And not just from that government: the big Chinese companies in the AI sector, such as Yitu, Megvii , SenseTime or the aforementioned CloudWalk (all valued at more than 1,000 million dollars ) also participate in this program.
According to Kai Fu-Lee , former president of Google China, the advantage of the Beijing regime when adopting new surveillance technologies is that, as a good dictatorship, they can afford to ignore “legal complexities” and “moral consensus” .