Google's Knowledge Graph isn't always right

Google Knowledge Graph blooper!

You've just heard of a new show and want to read what everyone is saying about it. What do you do? You head to Google and search for it. 

Way back in 2012 Google decided to supplement its regular results with the knowledge graph. The purpose was that "It provides structured and detailed information about the topic in addition to a list of links to other sites. The goal is that users would be able to use this information to resolve their query without having to navigate to other sites and assemble the information themselves."

Ok got it! Give you as much information as possible without you ever leaving the Google page. Sounds fantastic! While most of the time the knowledge panels that appear on the right seem to collate accurate information about a topic, sometimes you do see major goof ups like the one seen in the above screenshot.

Admittedly, "search party" is the kind of search (the irony!) phrase that would require a lot of sifting and disambiguation to arrive at the most relevant results for the query. Google has actually rightly guessed the intent, that is, information about the TV show that is rising in popularity, but the results returned seem to be a mixed bag. The traditional search results on the left are all about the new TV show, but the knowledge graph is a mix of the new show (2016) and an old one that aired in 2002. What gave it away (to me) almost immediately was the image used in the knowledge panel, the description i had read of the show definitely didn't indicate that it would be filled with buxom women and bare chested men. The inconsistency in the panel is rather silly once you realize what is happening, the IMDB rating is for the new show whereas the TV.com rating is for the old one, the first episode air date is for the new one whereas the episode and season information is for the old one..yeah you get the picture.

The pertinent question though is, does Google have human editors checking the knowledge panels for accuracy or are they depending too much on automation, algorithms and AI ?