Not a day goes by without noticing offensive posts and comments that are shared on social media. Moreover, the number of tragic victims of cyberbullying has increased immeasurably in the recent years. Therefore, social networking sites needed to be scanned for provocative words and posts. To tackle this menace, researchers have taken the advantage of artificial intelligence (AI).
It is a herculean task to filter out images that contain offensive words as it requires human involvement to look at each one. China’s censorship apparatus has found a new technique to root out the images containing banned words from social media apps and sites. According to SANS Internet Storm Center, an international monitoring organization, censors can scan images that are shared via China’s popular social media platform–WeChat.
Lately, words such as dictator, emperor, and throne were used to criticize Chinese President Xi Jinping, which are now banned in China. Several Chinese social media users tried to sneak in such words by featuring them in images. With the help of AI, these images can be scanned and removed. SANS Internet Storm Center tested their filter with 25 banned words, and were greeted with success. It tested the algorithm in a WeChat group with a few dozen people. The test was considered successful, as users in the U.S. received images containing banned words; however, the users of China did not come across such images.
Apart from WeChat, Instagram uses an advanced filter to block offensive comments. The main challenge in recognizing such content is a sentence structure. AI can learn to recognize such comments and ban them instantly. Facebook, launched a new AI, DeepText, which mimics the reasoning of human brain to understand languages, and filters out offensive posts and comments. When the system comes across a new word, it tries to deduce its meaning from the surrounding words. The algorithm, which has an error rate of just 1 out of 100 cases, helps AI to distinguish between the comments of friends and haters.
There are more than 2.8 million android apps on play store, and detecting the malicious apps is a tough challenge. Google has used AI to sort out the “bad apps” from its play store. Google claimed that the incorporation of AI has eliminated 99% malevolent apps containing insolent contents before anyone could install it. Apart from cyberbullying, scanning of content by AI can be used to mute unwanted texts, cursed words, and even spoilers of TV series. Twitter launched a similar feature, which enables user to block offensive content, phrases, and even unwanted emojis.
The use of AI has accelerated the process of removal of malicious contents from social media sites. As a result, the leading companies including Facebook, WeChat, and Google are investing their time and money against cyberbullying, especially since the problem is going to get worse soon. However, the time tell the future of the fight against cyberbullying.