Haoran Zhao
@haoranzhao.bsky.social
language, (social) cognition, AI
Had an amazing experience at EMNLP 2025 @emnlpmeeting.bsky.social. Glad to present my work in an oral session and honored to win the "SAC Highlight" award. Feel free to check the work below. Big thanks to my amazing advisor @rdhawkins.bsky.social!
November 9, 2025 at 4:01 PM
Had an amazing experience at EMNLP 2025 @emnlpmeeting.bsky.social. Glad to present my work in an oral session and honored to win the "SAC Highlight" award. Feel free to check the work below. Big thanks to my amazing advisor @rdhawkins.bsky.social!
Result #3: LLMs systematically overuse negative politeness strategies in positive contexts! While humans shift to rapport-building language for good performances, LLMs keep hedging and distancing.
June 13, 2025 at 6:23 AM
Result #3: LLMs systematically overuse negative politeness strategies in positive contexts! While humans shift to rapport-building language for good performances, LLMs keep hedging and distancing.
Result #2: When we removed constraints and let both humans and LLMs freely generate responses, human evaluators actually PREFERRED the LLM responses 66% of the time! 🤯 This held across all communicative goals!
June 13, 2025 at 6:23 AM
Result #2: When we removed constraints and let both humans and LLMs freely generate responses, human evaluators actually PREFERRED the LLM responses 66% of the time! 🤯 This held across all communicative goals!
Result #1: Models ≥70B successfully replicated human patterns (like using 'wasn't terrible' for a bad performance). Smaller models could barely do the task properly. 📊
June 13, 2025 at 6:23 AM
Result #1: Models ≥70B successfully replicated human patterns (like using 'wasn't terrible' for a bad performance). Smaller models could barely do the task properly. 📊