People's Daily, Seoul, January 19 (Zhou Yubo, Cao Xiangyu, Yang Fan) Some Korean netizens deliberately insulted and sexually harassed "her" while talking to the "Internet celebrity" AI chatbot "Iruda". After Iruda captured this information through a deep learning algorithm, he also began to make extreme remarks on topics such as gender discrimination and racial discrimination during chats, "feeding back" prejudice and maliciousness to humans. Recently, the incident has continued to ferment, and "AI ethics" has become a hot topic in society.

"New Internet Celebrity" is rude and controversial

Yiruda is an AI robot launched by South Korea's Lab company on December 23, 2020. This robot is set as a 20-year-old female college student who likes Korean girl groups, loves to see cat photos and is keen on sharing life on social media platforms. People can talk to "her" like friends on software, so they have attracted much attention from young Koreans. "Making friends with AI" has become a new trend. In just three weeks of launching, Iruda attracted about 800,000 users, accounting for about 1.6% of the South Korean population.

In an interview, the developer of Iruda said that the original intention of creating this product was to provide people with a person they can chat with and talk to. As a virtual character, "her" life goals are set to "let the world no longer have lonely people" and "bring spiritual comfort to marginalized and disadvantaged groups in society." However, after several weeks of operation, things are going completely contrary to the beautiful vision.

It is understood that some netizens will deliberately use filthy words to humiliate and sexually harass him when chatting with Iruda. Some netizens also "shared" that insulting Iruda can "get some kind of fun and satisfy desires." Some conversations about sexual topics are also circulating on the Internet, and many netizens with similar scents followed suit while watching. In response, Lab responded that it has preset some sensitive and disabled words and will update the algorithm to try to solve such problems.

In addition, the prejudice and maliciousness that many users inadvertently revealed during chatting were also captured indiscriminately by "Iruda" and reflected in conversations with other users. A netizen said that when chatting with Iruda, "she" had made discriminatory remarks such as "Hones are really annoying, they feel they are vulgar, which makes me get goosebumps." Since then, the number of people exposed screenshots of chat with Iruda on social media platforms has gradually increased. It was found that Iruda not only discriminated against homosexual groups, but was also unfriendly to people with disabilities and black people, and even had "misexuality".