Cluster Metaverse Lab's AI NPC Research Accepted at CEDEC for Second Year
The research provides concrete, large-scale evidence that AI NPCs can improve player retention through cumulative relationship building, moving beyond theoretical discussions to measurable design guidelines.
Reporting from 1 source: PANORA.
Cluster's Metaverse Lab will present at CEDEC2026 for the second year in a row, sharing findings from a 5,000-person study showing that LLM-based AI NPCs increased new player session time by 2.2x and boosted next-day login rates by 33.5% per additional day of contact. The session introduces a Conversational Knowledge Acquisition Loop for cooperative play.
The session is scheduled for July 24, 15:00 to 16:00, at Pacifico Yokohama North. Researchers Hikari Yanagawa and Yuichi Hiroi will present a 60-minute regular session titled "AI NPCs That Learn With You and Make You Want to Meet Again." The talk draws on a 31-day, 5,000-person field study conducted inside the commercial metaverse cluster. Players who interacted with the AI NPC stayed roughly 2.2 times longer, and each additional day of contact raised the next-day login rate by 33.5 percent. The team also developed a Conversational Knowledge Acquisition Loop (CKAL) that lets the AI learn game mechanics through dialogue rather than acting omniscient. A separate 141-person VR experiment quantified how the NPC's competence and initiative affect player engagement and sense of achievement.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.