Deep Dives Level Up Newsletters Saved Articles Challenges


What’s the deal with China’s new AI news anchor?

By Yair Oded

Nov 30, 2018


If you thought that Fox News is as scary as news channels can get, you better think again. Earlier this month, Chinese news agency Xinhua aired its (and the world’s) first ‘AI’ news anchor, a digital version of real-world Xinhua news anchor Qiu Hao. “This is my very first day at Xinhua News Agency,” declared the impeccably dressed artificial news anchor, who then congratulated himself for his ability to “tirelessly” deliver news, in both English and Chinese, 24/7. But as viewers across China, as well as around the world, eye Digital Hao with awe, some have already questioned whether he truly constitutes an AI breakthrough and whether his emergence means good or bad news for the world of media.

Xinhua’s anchor was developed through machine learning that imitates the voice, facial expressions and gestures of real-life broadcasters in order to forge what the company defines as “a lifelike image instead of a cold robot.” Yet despite Xinhua’s best efforts to simulate an actual human, many complain that the viewing experience of the ‘AI’ anchor isn’t so pleasing due to his flat, arrhythmic, and monotonic delivery.

His questionable imitation of human expression isn’t the digital anchor’s only problem. As pointed out by Will Knight, a senior editor for AI at MIT Technology Review, Xinhua’s virtual anchor is hardly an example of true AI. Knight argues that while machine learning enabled the news agency to come very close to flawlessly mimicking a real-life news persona, they still must feed the text to his virtual alter ego. “We should… always be really careful I think about the use of the term AI, and in this context you don’t want to suggest that this anchor is actually exhibiting any intelligence, because it’s not, it’s just like a kind of very sophisticated digital puppet,” Knight said to CNBC, “What actually creates those images and the movement of the lips and the voice of this anchor is using algorithms that are related to artificial intelligence. But to call this an AI anchor is slightly overselling it.”

Other than misrepresenting the field of AI and diverting the public further away from the true core and purpose of its research, Xinhua’s brand of virtual news anchors risks diminishing the quality of the news itself, seeing as the ‘person’ delivering them is incapable of intelligent analysis. Seeing as most media and news agencies adhere to some worldview, their anchors often deliver the content through the filter of their socio-political affiliation. And while this practice flavours news with a considerable degree of bias, it also helps the viewer engage in a more profound analysis of the events, as opposed to mindlessly digesting them as passive consumers of information.

News anchors draw our focus to particular issues as they surface and shed light on aspects of them we wouldn’t otherwise consider; they have the power to humanise stories and situate them within a broader context. But Digital Hao and his fellow virtualites stifle this type of analysis of news and essentially grant those behind the scenes who feed them the information absolute power over crafting the content viewers are exposed to without owing to the responsibility of actually delivering it. Then again, for the Chinese Communist Party this is of course a desirable outcome. And although in the West freedom of speech is still not as brutally infringed upon as in China, we’d be foolish to think that there aren’t characters in power here who are drooling at the thought of gaining full control over our intake of news.

Knight and his fellows bash Xinhua’s virtual anchor for being a sham AI, incapable of real intelligence and a servant of autocrats who wish to limit any type of unwanted analysis of the news. While their claims are legit, wouldn’t the alternative be even more terrifying? What if we finally managed to manufacture a computer programme actually capable of intelligently analysing written text and such technology was utilised to deliver news to the masses? Could true AI news anchors be trusted to exercise sound judgement, honesty and transparency as they narrate our societies’ stories? This may still be far down the road, but it is safe to predict that a digital ‘24/7’ Jeanine Pirro would mark the beginning of the apocalypse.