
Working with AI, keeping your judgment
与AI一起工作,保留自己的判断
这篇你会学到
At four in the afternoon, Mira has a clean draft and an uncomfortable question. The fictional editor is preparing a feature about a company’s new customer-support assistant. An AI tool has arranged her notes into brisk paragraphs, supplied an elegant opening and described the system as a success. The sentences are easier to read than her notes. They also sound more certain. Nothing in the polish tells her whether that certainty is warranted.
下午四点,米拉拿到了一份清爽的初稿,却遇到了一个让人不安的问题。这位虚构的编辑正在写一篇特写,主题是一家公司新推出的客服助手。AI工具把她的笔记整理成利落的段落,写了一个漂亮的开头,还把这套系统描述为成功案例。这些句子比她的笔记好读,也显得更有把握。然而,文字润色得再好,也无法告诉她,这种确定的口吻是否有充分依据。
Her dilemma is ordinary enough to matter beyond publishing. A support worker can produce a courteous reply that promises something the company cannot deliver; an analyst can generate an impressive explanation of a pattern that the data do not establish. In each case, the immediate output looks complete while a consequential decision remains unresolved. Working well with AI begins by noticing this gap between finishing a document and finishing the thinking that makes it defensible.
她的两难很常见,因此意义不只限于出版行业。客服人员可以写出礼貌的回复,却承诺了公司无法兑现的事;分析师可以对某个规律给出令人印象深刻的解释,但数据并不足以证明这一规律。两种情况下,眼前的产出看似完整,一个会带来重要后果的决定却仍悬而未决。要与AI配合好,首先得看见这道缝隙:写完文档,与完成足以让文档站得住脚的思考,并不是同一回事。