Google DeepMind：训练模型写道德论证更果断但判断力未提升
New Google DeepMind paper finds that training a model to write better moral arguments makes it more decisive, but no better at moral judgment.
Google DeepMind 新论文发现，用强化学习训练 Gemma-4 4B 撰写道德论证后，模型在 MoralChoice 模糊案例上因选项顺序调换而改变答案的比例从 14.3% 降至 6.6%，但附和用户的比例仅从 45.3% 变为 44.0%，道德寓言题准确率反而从 56.5% 降至 53.4%。论文建议评估建议或判断类任务时，将一致性与正确性分开测试。
正文
New Google DeepMind paper finds that training a model to write better moral arguments makes it more decisive, but no better at moral judgment.
LLMs are more and more often asked for moral advice. A model that keeps its answer when the options are reordered is often taken as a sign of sound judgment.
They trained Gemma-4 4B with reinforcement learning to argue dilemmas taken from classic novels. The model judged its own arguments in pairs, and the winners set the reward.
Answer flips when options were swapped fell from 14.3% to 6.6% on ambiguous MoralChoice cases. Yet siding with the user barely moved, from 45.3% to 44.0%, and accuracy on moral fables dropped from 56.5% to 53.4%.
If you evaluate models for advice or judgment tasks, test consistency and correctness separately, since a steadier model can still be wrong.