From text-completion algorithms writing scripts to machine learning models that spew fake news, Artificial intelligence is making all the headlines these days and is being used in all sorts of capacities.
Now, Kelsey Bressler, who is a developer, is helping her friend out in making an artificial intelligence system that detects and removes penis pictures sent via direct messages on Twitter (via BBC). The motivation for the project came after Ms.Bressler was sent an unsolicited penis picture on Twitter by a man. The developer said:
When you receive a photo unsolicited it feels disrespectful and violating. It's the virtual equivalent of flashing someone in the street. You're not giving them a chance to consent, you are forcing the image on them, and that is never OK.
She also remarked that the traditional way of blocking people sending such content should not be the way to go about dealing with cyber-flashing:
People say the easy solution is to close your messages - but we should not have to shut ourselves off to the world because some people do not know how to behave.
Ms. Bressler outlined the progress her team had made so far with regards to the software. They began with creating a Twitter account that was going to be used as bait and invited men to send in pictures of their genitals via direct messages. The team garnered a good response and the received samples were then run through the model to train it.
While the model struggled to flag photos of a man wearing a cage over his penis, initial reports seemed promising and the team reported success with detecting 'vanilla' penis pictures. The team has plans to share its research with Twitter upon the completion of the software.
16 Comments - Add comment