Silicon dreamin’
硅之梦
AI models make stuff up. How can hallucinations be controlled?
人工智能模型有时胡言乱语。如何控制幻觉？
- hallucination 幻觉，幻视，幻听（尤指生病或毒品所致）；幻视（或幻听）到的东西，幻觉，幻象
It is hard to do so without also limiting models’ power
不限制模型的能力很难做到
IT IS AN increasingly familiar experience. A request for help to a large language model (LLM) such as OpenAI’s ChatGPT is promptly met by a response that is confident, coherent and just plain wrong. In an AI model, such tendencies are usually described as hallucinations. A more informal word exists, however: these are the qualities of a great bullshitter.
我们已经越来越熟悉这样的情形。向OpenAI的ChatGPT这样的大语言模型提问时，模型会马上给出言之凿凿、流畅连贯却完全错误的回复。在AI模型中，这种倾向通常被称为幻觉。不过还有一个更通俗的说法：其实就是胡说八道，满嘴放炮。
- promptly 迅速地、立即；及时地、准时地；立即、马上
- coherent 喻指话语、观点、思想等“条理清楚的、明白易懂的、前后一致的”，强调逻辑清晰、组织良好，以自然或合理的方式连接、接续或整合，从而易于理解和说明（参见：小词详解 | coherent）
There are kinder ways to put it. In its instructions to users, OpenAI warns that ChatGPT “can make mistakes”. Anthropic, an American AI company, says that its LLM Claude “may display incorrect or harmful information”; Google’s Gemini warns users to “double-check its responses”. The throughline is this: no matter how fluent and confident AI-generated text sounds, it still cannot be trusted.
当然，也有更委婉的说法。OpenAI在给用户的说明中警告ChatGPT“可能会出错”。美国AI公司Anthropic表示，其大语言模型Claude“可能会展示不正确或有害的信息”，谷歌的Gemini则提醒用户“要复核它给出的回复”。总的主题就是，无论AI生成的文本有多流利自信，你都不能轻易相信。
