The tech giant uses complex artificial intelligence and machine learning to follow and analyze user activities, especially on its video hub, YouTube.
How do they do it? And should you feel threatened? Let’s dig deeper.
How does YouTube learn and remember my behavior?
Data is collected about your behavior as you navigate YouTube, Gmail, and other Google platforms, if you’re logged into your Google account. Website visits are also followed if you log into the Chrome browser with your Google account. Your searches, preferences, and habits are tracked and used to predict what you might do next.
Google is also trying to “teach” YouTube artificial intelligence algorithms how humans interact by showing it 57,600 videos of people sitting, drinking beer, talking, hugging, and other human activities. This will also teach the program what how to anticipate next moves from human behavior.
How is YouTube using AI?
YouTube uses all this data to create algorithms that acquire insight into which videos it recommends to its visitors. Google Brain supplies supporting data to give more in-depth knowledge and tracking for viewing recommendations.
YouTube also uses artificial intelligence to track and identify copyright infringement in its videos. In this technology, called Content ID, content is scanned and compared against media already stored in the YouTube database.
Who is YouTube’s AI focused on?
YouTube focuses on individual video viewers, but increasingly is shifting its attention onto brands and companies to optimize its advertising.
Sponsored videos are marketed on Google’s search results and within the recommendation lists on other YouTube videos. This is an ever-increasing money-maker as YouTube looks to compete with other online TV streaming services such as Hulu, Netflix, and Amazon.
YouTube is also looking to use its software to identify and remove inappropriate videos, which can cause advertisers to think twice about spending ad dollars.
How is YouTube helping boost advertising dollars?
Because YouTube is owned by Google, the amount of data available is huge. All browser history, searches, mail preferences, and profile details (enter Google Plus if you have it), amount to trillions of data points that are used to create the perfect preference selections for advertisers.
The artificial intelligence program that YouTube uses helps it to identify not only who to best market to, but also how much to charge for reach and predictability in engagement and views. Each recommendation is personalized to each user, brand, and video content.
How does YouTube know what videos to recommend next?
For the everyday video viewer, YouTube recommends a series of automatically-generated, related videos in several locations — lists on the right and below the current video (on desktop), search results, and in the video viewer after the current video completes.
In 2016, YouTube released an update to its recommendation system to base it on a “deep neural network technology” that recommends “hundreds of millions of different videos, billions of times, in 76 languages.” According to Google, this tech can “find patterns automatically and keep learning and improving as it goes.”
All views are adapted to a user’s preferences and watch history, and they receive feedback based on if you click or not. Ultimately, the system itself could be providing valuable feedback to the company about what types of movies to purchase, feature, and license on their subscription stream.
Should I fear/allow this tracking?
It depends on how much information you want companies like Google to collect from you, and if you’re willing to allow this collection to improve your web experience. There are ways to limit “invasions of privacy” like user data collection: using a VPN to access the Internet, browsing in private or “incognito” mode, disabling cookies, just to name a few.