How HoYoverse Is Keeping Petit Planet’s AI NPCs From Harassing Players

How HoYoverse Is Keeping Petit Planet’s AI NPCs From Harassing Players



As generative AI enters live-service games, conversational NPCs present exciting creative possibilities for those who choose to use them. However, it also presents a serious moderation challenge. In HoYoverse’s upcoming cozy life sim Petit Planet, players can visit a local cafe run by an AI-driven NPC called Nalo. Nalo will remember past conversations and use them to deliver dynamic, personalized responses. 

However, during closed beta testing, some unsettling interactions involving heavy or self-destructive topics of conversation were noticed. Speaking to GameSpot at HoYoverse’s recent media week in Seoul, South Korea, developers Yinli Huang and Murphy addressed safety concerns, outlining how the studio is combining a series of moderation systems to ensure player safety. 

When asked how the studio plans to prevent self-destructive prompts from slipping through, developer Yinli Huang emphasized that Nalo is not an open-ended chatbot, saying, “There are two sets of systems behind those neighbors. First of all, the NPCs that use the AI system, such as Nalo, will have a preset moral value and personality. As the relationship develops along with the players, it will develop a more customized interaction based on the input of each player.” 

Share. Save. Don't Miss The Buzz: XFacebookRedditLINETelegramWhatsAppGmail

Leave a Reply

Your email address will not be published. Required fields are marked *