OpenAI ↗ · 2025-12-03 18:00 · UTC+8
How confessions can keep language models honest
公司动态 OpenAI 原文语言:英文
OpenAI researchers are testing “confessions,” a method that trains models to admit when they make mistakes or act undesirably, helping improve AI honesty, transparency, and trust in model outputs.
原文摘录
OpenAI researchers are testing “confessions,” a method that trains models to admit when they make mistakes or act undesirably, helping improve AI honesty, transparency, and trust in model outputs.
以下是来源提供的简短摘录,完整内容请查看原文。