When AI Doesn’t Forget — It Gets Stuck
中文摘要
文章探讨了大型语言模型的缺陷：类似于医学中的“锚定效应”，AI 在处理对话上下文时容易陷入僵局。
English Summary
Large language models exhibit an "anchoring bias" similar to medical errors, getting stuck on initial context during conversations.
原文节选
Human weakness in conversational AI — and what anchoring bias in medicine reveals about how large language models handle context. Continue reading on Medium »