Standards for assessing gender violence risks must be established from the beginning of AI development.

three line summary

The Korea Women's Policy Institute announced on the 23rd that standards should be established to check the risk of gender violence from the stage of developing and evaluating artificial intelligence services.

The researcher pointed out that while the United States blocks data learning by classifying profanity, hate expressions, and sexually explicit or violent information as “harmful content,” Korea simply enters a prompt that states, “Do not give sexist or racist answers.”

The researcher emphasized, "To prevent AI technology from being used as a tool to spread digital sex crimes, safety standards must be specified to check and respond to risks in advance."

We use ‘three-line summary’ technology, which is automatically shortened by artificial intelligence.

To understand the full story, you should read it in conjunction with the main text of the article.

(Seoul = Yonhap News) Reporter Hong Jun-seok = The Korea Women's Policy Institute announced on the 23rd that standards should be established to check the risk of gender violence from the stage of developing and evaluating artificial intelligence (AI) services.

The researcher made this suggestion, saying that as a result of an in-depth interview with eight domestic experts in charge of AI service development, it was confirmed that awareness of AI ethics and safety takes a back seat compared to performance and profits.

The researcher pointed out that while the United States blocks data learning by classifying profanity, hate expressions, and sexually explicit or violent information as “harmful content,” Korea simply enters a prompt that states, “Do not give sexist or racist answers.”

The researcher emphasized, "To prevent AI technology from being used as a tool to spread digital sex crimes, safety standards must be specified to check and respond to risks in advance."

Report: KakaoTalk okjebo <Copyright (c) Yonhap News, unauthorized reproduction and redistribution, AI learning and use prohibited>