Porn is going to get a lot smarter. Well, not porn exactly, but ID-ing it and identifying the performers in it, all for better and improved indexing so searching for specific types and categories of porn will be easier.
Pornhub Searching To Be Easier Thanks To Machine Learning
Pornhub is at the forefront of this porn searching evolution, announcing on Wednesday, Oct. 11, that it's using machine learning to automatically catalog its huge library of adult videos.
The company is going to start small, employing facial recognition technology to detect 10,000 different adult performers and tag them individually in videos they appear in. As The Verge notes, this type of information is usually provided by publishers and viewers, who will remain crucial to the whole process by confirming whether or not facial recognition tagged the correct performers.
Pornhub plans to scan all 5 million videos "within the next year" and then shift toward more complex methods of identification, like determining where videos should be categorized in, tagging them as "public" or "blonde," for instance. This is presumably more complicated because Pornhub's machine learning technology has to actually figure out what the context is apart from who the performers starring in a video are.
Corey Price, Pronhub's VP, said in a press statement that the company is jumping on the artificial intelligence bandwagon to "expedite antiquated processes." But how fast will it be, really? While in beta, Pornhub's machine learning software was apparently only able to scan 50,000 videos in a month. If the site's gigantic library of adult content is factored in, tagging them all would probably take many, many years.
But apparently, the goal here isn't to categorize or tag every single video but to make searching for porn as easy as possible for viewers, and as Price says, the site mainly wants to provide users whatever type of video they're looking for quickly.
"The easier it is to navigate and access what they want, the happier they will be," he said.
Here's Why This Might Be A Bad Thing
The implementation of machine learning, facial recognition, and AI on a porn site seems simple and harmless enough on the surface level, but it actually has concerning implications for privacy, as The Verge points out. For one, there's the problem of revenge porn often appearing on popular porn sites, so will Pornhub tag the people in those too? What if an adult performer chooses to stay anonymous? What system does Pornhub have in place for those who don't consent to being identified?
The company claims that its machine learning software will only be limited to identifying performers that's already in its database. That being said, the idea that performers can be identified via software is still disconcerting, although it remains to be seen just how well Pornhub's software works.
