LLMs lose reasoning ability during emotionally complex conversations, French researchers find
Original article in Korean"인간 공감해주다 패닉? AI 스트레스 받으면 답변 붕괴"...프랑스 연구팀 규명
Korean outletAI Times
Summary
Researchers from France's Ryerson Research Institute published a study on algorithmic emotional blunting in large language models. The paper, published in Springer Nature's Discover Artificial Intelligence journal, shows that LLMs' interpretation ability degrades sharply when handling emotionally complex dialogue.
aitimes.com · in Korean