“AI Anchor” to read headlines for Chinese state-run news agency

Earlier today at an internet conference in Wuzhen, it was announced that China’s state-run news station, Xinhua, will be using “Ai anchors” in the future to read the news. The AI-anchor technology was developed in collaboration with Sogou, a local Chinese search engine that has been doing research in the field of AI for quite some time. These ‘anchor-bots’ are digital melds made from footage of human anchors who use synthetic voices to read the news.

While the government has not released details on the tech used to create the anchors, they’re supposedly up-to-date with current machine learning research. Xinhua appears to use human anchors as a base layer for their ‘anchor-bot’ and then digital parts of the mouth that allow the speaker to be controlled. Through the use of a synthetic voice, Xinhua’s anchors can read the news at a faster rate than traditional CGI.

South China Morning Post reports that two separate anchors have been created for English-language broadcasts and Chinese-language ones. The anchors reportedly provide a cheaper option for the agency to distribute news reports for their TV, online, and WeChat mobile output. Xinhua’s representatives claim there are “endless” prospects for the device.

According to Xinhua, all of its AI anchors “work 24 hours a day on its official website and various social media platforms, reducing news production costs and improving efficiency.”

If the tech is successful, it may undercut the use of human anchors and therefore reduce the value of their employment. In countries like the US, Britain, or Japan, news-anchors tend to be valued and well-paid by their employers. While Xinhua’s state anchors are paid substantially less (even when accounting for purchasing parity), they still stand to receive endorsement deals because of the publicity they receive through their job. If AI replacements become the new norm, they might become obsolete.

However, this worry may be premature given the state of the AI’s current set of abilities. In addition, the tech still does seem to have some issues. While it looks like an interesting novelty, its blatantly obvious from the video above of the English-language ‘anchor-bot’ that its voice is artificial and its facial expressions have limited capabilities.

In order to alleviate such concerns, Xinhua assures its followers that machine learning researchers are making swift improvements to develop in this respect, and soon enough viewers will not be able to tell if someone is a human or AI anchor.

These changes to how news stories are delivered in the media might excite most tech aficionados, however, there are some worries about such technology being used to more effectively spread government propaganda. Considering their extensive press censorship, it remains impossible to receive unfiltered reports of controversial domestic events, such as the government’s suppression of Tibetan or Muslim Uyghur minority communities. Some fear that AI anchors might heighten an already hyper-controlled system of news distribution.

But in reality, it’s difficult to judge if these Ai anchors will have any profound impacts on wider society. Even prior to the development of ‘anchor-bot’ technology, the government was still able to control how the anchors read the news without protest. In addition, synthetic figures are slowly integrating themselves into popular culture, through popstars like Hatsune Miku and CGI Instagram models like Lil Maquela. AI anchors may just be an extension of an increasingly AI-dominated society.