Are Xinhua’s News Anchors Really AIs?

If you’ve been keeping up with my recent posts, then you will remember us mentioning the AI news anchors that seem to have debuted in China. As we stated before, even though they depend on supervised learning principles to function, their creation may still be considered a breakthrough for the news industry.

One major reason this could be true is that having AI news anchors could relieve some of the pressure on human news anchors in terms of reducing their workload. More specifically, since these AIs are supposed to be able to work around the clock, perhaps most news anchors would be able to go home to their families on a more regular basis.

In reality, however, none of this seems to quite be the case for two principal reasons.

Firstly, since these AIs use supervised learning principles to run, some sort of human worker has to continuously feed them data that they then report on. Given this, it would seem to be certain that we have not reached a time in which AIs can truly take our jobs, at least not in this context.

Secondly, and more recently, certain industry experts have begun to cast doubt on whether these two inventions are AIs at all, or something else entirely. Given the fact that the principle mention of this comes from CNBC, a popular news media outlet, it does appear logical to take it with a grain of salt.

On the other hand, the key sources that the CNBC article uses to substantiate its claims appear quite trustworthy in terms of their experience in the AI industry. Central to this group is MIT, which many now think of as not just the world’s best technological university, but also the world’s best blockchain and AI research center.

Furthermore, the expert from MIT who is quoted in this particular article is Will Knight, who works as one of MIT’s senior AI editors for its’ news wing.

Overarching his criticisms of the two AIs is the idea that using the term AI here might be a misnomer, because these particular systems are not capable of demonstrating any true intelligence alone.

If you’re wondering exactly what he means by this, it appears that Knight was attempting to shift our focus to how we can really understand these special news anchors. In other words, if AI is not the right term for them, then what is?

In a nutshell, what he concluded was that it would be better to call them Machine Learning systems which have the built-in skill of learning how to do the job of a human news anchor.

To this, Knight added that for them to be AIs, they would have to demonstrate their own ingenuinity in terms of something like creating and presenting a news script on their own. If you are acquainted with the basic definition of an AI system, then Knight’s claims will likely ring true. Also, related to this, there have been plenty of proven instances in which Machine Learning algorithms are used outside of AI systems.

AIs are computer systems, built with the help of Machine Learning, that can take in data and use that data to perform some sort of intelligent action more efficiently over time. Because of this ability, they are made to create, not just react.

Until Xinhua’s anchors demonstrate such a skill, it seems reasonable to conclude that we actually might not truly have AI anchors just yet. In future pieces, expect us to dig further into this and related topics to illustrate the differences between simple Machine Learning systems and true AIs.