arXivDennis Fucci, Andrea Bacciu, Dong Liu, Weronika Łajewska, Saab MansourTue, Oct 6, 2026, 8:17 AM PDT
score 16.6
Real Wikipedia chats used to test multilingual AI personas
Original: Wiki-Talkie: Multilingual Benchmarking of Persona-Based Agents on Real-World Discussions
Source: arxiv.org ↗
Writing ELI5 summary…