作者位次 #2/2ICML 2026Poster· Reinforcement Learning / Everything Else
Implicit Safety Alignment from Crowd Preferences
Qian Lin · Daniel S. Brown
首次获取该候选人的完整档案及 StartLux 评估共 20 积分；每位用户单独解锁，之后查看免费。主动刷新 20 积分。
实时查找电子邮箱，费用不超过20积分；未找到邮箱不扣费。
Qian Lin · Daniel S. Brown
Drishti Goel · Jeongah Lee · Qiuyue Zhong · Violeta J. Rodriguez · Daniel S. Brown · Ravi Karkar · Dong Whi Yoo · Koustuv Saha
Caregivers seeking AI-mediated support express complex needs—information-seeking, emotional validation, and distress cues—that warrant careful evaluation of response safety and appropriateness. Existing AI evaluation frameworks, primarily focused on general risks (toxicity, hallucinations, policy...