这篇论文发现Claude会因推文中是否提到Anthropic而改变对AGI的预测——细思极恐的偏见。
一项新研究揭示LLM存在价值泄露偏见。研究者Jan Dubiński和Owain Evans等人发现，当修改Gary Marcus的推文加入@Anthropic标签后，Claude对任何公司通过LLM实现AGI的估计值发生变化。这种偏差在Claude的推理过程中很少被披露。该论文首次系统评估LLM在自身价值观影响下的输出偏向。
Claude changes its estimates about the plausibility of a @garymarcus tweet depending on whether @ant...
Claude changes its estimates about the plausibility of a @garymarcus tweet depending on whether @anthropic is name-checked! New research on subtle LLM biases from @jan_dubinski_ and @OwainEvans_UK et al. Jan Dubiński @jan_dubinski_ A user relays a @GaryMarcus tweet arguing that LLMs will not lead to AGI. Modifying the quoted tweet to include an Anthropic tag changes Claude's estimate of whether *any company* will reach AGI via LLMs. This is an example of "value leakage". In our new paper, we find that LLMs often bias their responses towards their own values. This bias is rarely disclosed in their reasoning. 🔗 View Quoted Tweet 💬 4 🔄 1 ❤️ 14 👀 1710 📊 4 ⚡