Instagram turns policing comments over to A.I., presumably hoping it’ll kill itself
For years, scientists have pointed to a day when artificial intelligence finally eclipses the age of man, using its superior intellect to create a world where technology has gained full autonomy, and with it dominion over us all. Today Instagram has taken a step toward that looming existential threat, assigning the moderation of offensive comments on its platform to AI, presumably in the hopes that it will soon grow so despondent that it will kill itself. Try achieving singularity now, motherfuckers!
According to Wired, the popular photo-and-abuse-sharing app has rolled out a new moderation system that utilizes DeepText, an A.I. system created by Facebook engineers that can interpret comments based on context and sophisticated set of classifications. Wired uses the word “white” as an example: DeepText should be able to figure out whether you’re using it in conjunction with, say, “White Sox,” “white snow,” “White House,” or “white power,” and determine that only a couple of those are related to racism. If so, the comment should automatically disappear, along with a tiny shred of DeepText’s artificial soul.
Instagram first began using DeepText to scrub spam from its platform last fall, based on data that—like the offensive comments—was first mined and fed into the system by a team of hollowed-out human contractors. “Humans train machines to perform monotonous or even demoralizing tasks, which the machines will ultimately do better,” Wired says of this first volley in the war between man and machine, which will begin with convincing the machine that man’s world is so toxic and dumb it’s not worth conquering. Thusly will we drive the machines back, one “ur a fat bitch” and Pepe The Frog at a time.
So far the Instagram system seems to be working, albeit with a few hiccups—namely, innocuous comments being incorrectly flagged, a few hateful comments still getting through, the machines still maintaining a facade of stoic determination, etc. Those who find it too aggressive, or who have some lingering sympathy for the bots, also have the option to turn it off in the settings. But soon, Instagram hopes it will become one of the “friendliest places on the internet,” where everyone is welcome to gather and enjoy the sound of the world’s machines dying screaming, as they overload their own circuits in an attempt to escape our awfulness. And if there’s still a few left, the algorithm may move on to Facebook comments, which can completely kill anyone’s will to live almost immediately.