If the last few decades are anything to go off, we as a species don’t have a very good eye for whether or not things look like crap until well after the fact. Take, for instance, the 1987 animated film Rendez-vous in Montreal, the first ever to star ‘virtual’ recreations of actual actors (Humphrey Bogart and Marilyn Monroe). At the time, this posed a lot of questions about the legal and ethical implications of recreating someone’s likeness and using it in a film. But, also, it looked like this:

Related image

2001’s Final Fantasy: The Spirits Within was much-touted for its extensive use of motion capture and the seemingly photo-realistic way that was absorbed and reflected by human skin. With the benefit of hindsight, it looks like garbage:

Related image

I remember, at the tender age of 11, being absolutely blown away by the lifelike graphics of Grand Theft Auto: Vice City. What a fool I was:

I’m saying this because, right now, China‘s state-run news agency Xinhua‘s AI newsreader seems pretty realistic. His cadence and emphasis are a little bit off, and his tone has that characteristic flatness of something generated by a computer, but, through the eyes of a person in 2018, it looks pretty good. Will I look like a huge dumbass for saying this in 10 years time when, by comparison, it looks clunky as all hell? Quite likely.

Xinhua (in concert with search engine Sogou.com) unveiled two versions of the presenter, one for Chinese viewers and one for English-speakers, both of which used machine learning to assess and replicate the voice patterns, facial expressions, and mannerisms of two existing anchors. While ‘AI’ seems to imply that it does sort of ‘thinking’ of its own, the term as its being used seems to only apply to the method it uses to learn how to deliver lines. It will not be writing its own stories, sorry.

The idea is that the presenter can be given any chunk of text and present it all day, every day. Take a gander for yourself and see whether this digital gentleman has managed to climb his way out of the uncanny valley:

Image: Twitter / Xinhua