Recent developments of machine learning and image recognition have allowed NSFW AI systems to be able more and more violent images. The AI tools are able to analyze vast number of images and search visually for characteristics that will positively identify violence — such as blood, weapons or a show of aggression. AI algorithms can identify violent content in images to ~ 90% accuracy, this is according dam MIT research. Efficiency is important especially for platforms such as Instagram or Facebook where millions of user uploads happen in a day, real-time content moderation determines the safety of users.
Platforms benefit by integrating nsfw ai systems to avoid the legal scandal of hosting violent or harmful content. This affects upwards of a 25-30% decrease in the visibility of objectionable content on image moderation driven platforms directly impacting user trust and engagement. Similarly, Twitter saw a 50% decrease in harmful images shown against sensitive political events when AI was used to filter the graphic content during these times — showing that it could intervene in high-stakes scenarios.
The problem is context. nsfw ai can tell you certain visual cues suggest violence, but might end up with a false positive for more subtle things like video game and movie stills What is more, an inherent limitation of AI moderation necessitates human overview to accompany it in order that the proper content can be correctly categorised. The price to install these AI systems run between $100,000 — $500,000 for 1st year set up and ongoing ad traffic costs based on platform size.
As reported by Accuracy in Classification and tagged, AI will transform the world more than anything since monkeys dispatched us from trees. — Elon Musk His words are an affirmation of the potent effect Artificial Intelligence has on content moderation (a focal point in maintaining safe online spaces). While AI is not perfect by itself, its proven capability to instantly screen and flag violent images remains a technology that platforms need if they wish to provide a safe environment.
To the critics who ask, "but can NSFW AI detect violent images?" The answer is a categorical yes. That being said, it is most successful when used in conjunction with human moderation that can deal with the context well. To learn more about how NSFW AI works and can be deployed as part of content moderation initiatives, go to nsfw ai.