Why do US AIs (ChatGPT, Gemini) become "cowardly" and ambiguous on scientific and political issues, while models from other countries give direct answers?
OpenAI 相关模型动态已经出现，适合跟踪能力变化、生态影响和后续可用性。
作者抱怨美系AI在气候变化、疫苗、压裂危害等共识议题上常以“有人认为…很复杂”等套话模糊其词，仿佛把既定科学当成政见之争；而他测到的他国模型更直接、基于证据。其假设是美企为规避被起诉或被各派抵制、避免被指偏见，宁可两边摆平，造成把事实与谬误等量齐观。这会
I have noticed a frustrating pattern and I want to know if anyone else is experiencing it. When I ask models like ChatGPT or Gemini about topics with scientific consensus (such as climate change, vaccine efficacy, or the harms of fracking), the answers are extremely ambiguous. They use phrases like "some argue...", "it is a complex issue...", "there are various positions...", treating scientific facts as if they were debatable political opinions.
It's like asking "Is the Earth flat?" and having them reply: "There are those who say it is round and others who say it is flat, the truth depends on your perspective."
However, if I ask the same questions to models developed in other countries (outside the Silicon Valley ecosystem), the answers are direct, evidence-based, and straightforward. It's not that they are "left-wing," it's just that they present reality as the data has proven it.
My hypothesis: I think US companies prioritize not being sued or boycotted by any political group over telling the truth. They are so afraid of being accused of "bias" that they end up validating misinformation by giving the same weight to a proven fact as to a lie.
This is dangerous. If my family uses these tools and the AI tells them that climate change is "an open debate," ignorance is reinforced. Neutrality should not apply to facts.