精选理由
Anthropic推迟开源MHS，称LLMs仍需提升物理直觉能力，研究预览版将强化AI物理世界应用的安全评估。
Anthropic表示在开源MHS模型前仍需更多研究。该模型目前缺乏物理直觉，仅通过文本和图像学习物理世界知识。研究预览版将帮助构建更多安全评估，加强AI在物理世界应用的保护措施。
原文 · Anthropic
There’s more to learn before we open source MHS. LLMs still lack physical intuition, having learned ...
There’s more to learn before we open source MHS. LLMs still lack physical intuition, having learned about the physical world from text and images. The research preview will let us build more safety evaluations and strengthen protections for using AI in the physical world. 💬 3 🔄 0 ❤️ 37 👀 2344 📊 5 ⚡