How does YouTube’s evil algorithm work?

YouTube’s automated video proposal system has been heavily criticized in the past for being able to produce videos of malicious themes on its own, even though users are not actively searching for them.

What is the purpose of the proposed system?

Google said its proposal system uses AI to suggest to users what works best for them. However, the fact that this system still has many problems. Guillaume Chaslot, a former Google employee who worked with the proposed algorithm, also pointed out this problem.

“It’s bad that YouTube uses AI again to suggest videos to you. If it works, it will help you find exactly what you want, which is really great. However, the problem is that artificial intelligence is not created to show you what you want, but for you to be addicted to YouTube. The proposed system is designed to consume users’ time, ”Chaslot told TNW.

Having worked at Google, Guillaume Chaslot created a tool for users to recognize how the social network’s proposed system works. Photo: AlgoTransparency.

According to Chaslot, the most important criterion for evaluating a “successful” proposition is time to see. The longer the viewing time, the more ads YouTube displays to users, but that’s not exactly what users want.

Because of YouTube’s emphasis on viewing time, content such as conspiracy theory, fake news will be proposed more often. Content like this is classified as “sensitive”, ie, does not violate YouTube’s policy but still contains offensive or offensive subjects specific objects.

For those who like to watch a particular topic, like music or games, the proposed system works fine. However, for those who consider YouTube as a news or news channel, the recommendations from this social network may spread false information, causing many to receive extreme information. The more content that is sensitive, controversial, the easier it is for a viewer to click and watch, and the more the system recommends.

“We must realize that YouTube’s proposal system is malicious and falsifies all arguments. Currently, this system promotes sensitive content, is not banned but attracts interest, “Chaslot said.

Look for solutions to overpower AI

Former Google employees have repeatedly voiced about the ultimate purpose of the YouTube suggestion system. He even created a tool called AlgoTransparency after leaving YouTube to show people what this social network suggests based on video.

AlgoTransparency shows the most shared videos on popular channels. When the special investigator Mueller published a conclusions investigating Russia’s involvement during the 2016 US presidential election, Chaslot pointed out that the video was shared by most YouTube channels during that time as a video. from RT, Russia’s national radio station.

According to former Google employees, YouTube’s algorithm has the ultimate purpose of enticing users to watch longer, so the more sensitive content is proposed. Photo: Reddit.

He pointed out that this video has a sudden increase in the number of proposals within 2 days, then almost no recommendation. Views of videos on RT channels are also quite low.

“It’s strange to see that there are so many channels that introduce this video, especially when considering its views. No one understands why the video was proposed, ”concludes Chaslot.

When asked, Google denied Chaslot’s explanation, and said they could not say what the YouTube proposal system was based on. TNW also tried to create a re-inspection system like AlgoTransparency, but could not have the same results as the Chaslot announced.

Chaslot said that to really test his words, Google must be more transparent about its proposal system. However, now YouTube owners say very little about this system.

Browser extensions such as Nudge can filter addictive content on social networks, but this is only a solution to the consequences, not the cause. Photo: TNW.

“The best short-term solution is to delete the proposed function. I don’t think this is a useful feature for users. If YouTube wants to suggest, they should email it or just suggest channels that users subscribe, ”Chaslot said.

In addition, the former Google employee also recommends using Chrome extension software with a Nudge name to remove addictive content on Facebook and YouTube. Of course, these practices only help eliminate consequences, not causes. Chaslot said that should focus on long-term solutions.

“In the long term, people need to control AI, not to control AI,” Chaslot said.

Make money from YouTube: Is it ‘delicious’? The trend of more and more YouTube users as a place to make money is growing. But is this really easy to make money?