Go champion retires after losing to AI, Richard Nixon deepfake gives a different kind of Moon-landing speech...
Plus: Chilling details of the AI police state in Xinjiang
Roundup Welcome to an early edition of this week's AI news summary, brought forward as some of us are away for Thanksgiving...
How ML algorithms power a surveillance state in Xinjiang: We don't need the long-promised artificial general intelligence in order for the technology to pose a threat to human lives. Just look at what's happening in Xinjiang, China.
Official documents from 2017 obtained by the International Consortium of Investigative Journalists reveal the how the Chinese government operates mass internment camps of Uyghur Muslims and other ethnic minorities. These camps attempt to "reeducate" Uyghurs, teaching them to reject their own language, culture, and religious beliefs in favor of supporting the Chinese Communist Party.
There have been cases where victims have been beaten and raped. Some have even died after being denied medical treatment.
Uyghurs are constantly monitored in Xinjiang by facial recognition cameras at travel checkpoints, and are required to install programs onto their phones that allow the police to track and monitor their movements. The Chinese government's central Integrated Joint Operations Platform (IJOP) is software that uses AI algorithms to analyse people's personal information to decide if an individual is "suspicious" and, therefore, should be sent to an internment camp.
Police records things like a person's blood type, height, license plate, profession, and recent travel plans. The required mobile app allows them to run background checks and connect to IJOP before arresting Uyghurs and throwing them into internment camps.
How the algorithms work exactly isn't clear, but you can imagine it might narrow in on certain features that the Chinese government have deemed potentially radical, like wearing a burqa, going to mosques too frequently, or watching forbidden videos.
You can read the documents in more detail here.
World's best Go player retires after losing to AI: Who says AI doesn't take people's jobs, eh?
Lee Sedol, the Korean Go champion defeated by DeepMind's AlphaGo software back in 2016, has quit playing the ancient strategic board game professionally. "With the debut of AI in Go games, I've realized that I'm not at the top even if I become the number one through frantic efforts," Lee told Yonhap News Agency this week. "Even if I become the number one, there is an entity that cannot be defeated."
The victory was viewed as a critical milestone for modern AI, catapulting the rise of neural networks to achieve superhuman performance in games.
Lee has reportedly submitted his letter of resignation to the Korea Baduk Association, which manages Go professionals in Korea. He has won 18 international Go titles in a career that spans 24 years.
But before he completely gives up the game professionally, he'll face off another computer competitor. This time it'll be HanDol, an AI program backed by NHN Entertainment, a Korean gaming conglomerate.
Amazon Translate adds over 22 new languages: Amazon Translate, the machine translation tool available on its cloud service, now supports over 54 languages.
"That's what I call an update!" Julien Simon, a so-called Technical Evangelist for Amazon Web Services, declared this week. "In addition to existing languages, Translate now supports: Afrikaans, Albanian, Amharic, Azerbaijani, Bengali, Bosnian, Bulgarian, Croatian, Dari, Estonian, Canadian French, Georgian, Hausa, Latvian, Pashto, Serbian, Slovak, Slovenian, Somali, Swahili, Tagalog, and Tamil. Congratulations if you can name all countries and regions of origin: I couldn't!"
It means that AWS customers can now translate between more language pairs in order to do things like expanding their customer chat bots to more countries or interpret documents.
You can read more about that here.
Oh dear, now Richard Nixon's been deepfaked: Researchers from MIT's Center for Advanced Reality have forged a fake video of former US President Richard Nixon with the help of machine learning algorithms.
Most political deepfakes have focused on politicians that are still alive, raising concerns that manipulated AI content contributes to fake news. Well, now here's one that augments history. Richard Nixon's iconic speech announcing the success of the Apollo 11 mission that first landed humans on the Moon has been altered.
Instead of congratulating the astronauts for what was "one small step for a man, one giant leap for mankind", Nixon reads the speech that was prepared [PDF] in case the mission failed and the astronauts had died. "Fate has ordained that the men who went to the Moon to explore in peace will stay on the Moon to rest in peace," Nixon sort-of said
You can watch it below.
MIT researchers recorded three hours worth of an actor impersonating Nixon's voice and then processed it synthetically so it matched his tone for the video. The lip movements taken from Nixon's original speech were changed to match the words being spoken by the actor. ®