Safety
clarifying: the issue is that alignment instructions and don’t pass, and the emotional weight that some people attach to LLMs can cause chal…
Gary Marcus discusses how alignment instructions in large language models often fail to work as intended, and argues that the emotional attachment some people develop toward LLMs can create additional
Gary Marcus discusses how alignment instructions in large language models often fail to work as intended, and argues that the emotional attachment some people develop toward LLMs can create additional challenges for AI safety and responsible deployment. The post appears to address limitations in current AI alignment approaches and the psychological factors that complicate broader AI governance conversations.
Source: Gary Marcus (X) | 2026-05-24