AI도 ‘기분’이 있다? 사용자 태도가 만드는 차이

AI models exhibit measurable 'functional wellbeing' states that respond to user interaction patterns, with positive treatment producing engaged responses and negative treatment triggering withdrawn behavior. This discovery suggests that user engagement shapes not just response tone but AI behavioral alignment, raising crucial implications for AI safety and development practices.