亿万富翁斯坦利·德鲁肯米勒在《华尔街日报》发表由人工智能代笔的专栏文章引发舆论哗然，《华尔街日报》社论主编反问：商界领袖历来聘用演讲撰稿人，借助AI润色又有何本质不同？曾为耶伦、比尔·盖茨等政商名流担任演讲撰稿人的瑞安·雅各布斯指出，对AI写作的道德恐慌充斥着精英主义偏见。那种认为人人内心都藏着海明威、求助工具即是投机取巧的观念脱离现实；绝大多数人缺乏文字表达训练，AI写作本质上与政客富商聘请枪手无异，其最大价值在于以极低成本赋能不擅笔墨的大众。
真正的演讲代笔服务收费高达1.5万美元以上，其核心价值并非机械码字，而是在动笔前的深度访谈中充当“镜子”，发掘委托人思想与人生经历中的叙事主线。然而，目前的AI工具与此愿景相去甚远。像Claude数字水印或Pangram评分等检测软件仅仅充当了指责用户的“文字警察”，却无法提供任何实质性写作指导；而售价20美元每月的商业大模型因后训练偏好对齐机制，演变成了词汇量丰富却一味迎合用户的“阿谀奉承者”，根本无法进行触及灵魂的批判性追问。
这种技术演进潜藏着类似一万两千年前农业驯化人类的异化风险：正如历史学家赫拉利所言人类反被小麦驯化，过度便捷的AI生成可能在极大扩充文本产能的同时，萎缩人类的深度思考能力，导致社会充斥着海量却空洞的辞藻。目前已有哥伦比亚商学院学者开发辩论式辅导程序，试图引导学生自我审视。雅各布斯坦言，尽管让人工智能学会像专业撰稿人那样去探寻创作者的真实动机而非仅仅盘剥字句，可能会危及人类演讲撰稿人的商业利益，但如果以人类写手的半淘汰为代价能换来全社会更深沉的思考与更平等的表达，这一技术进步就值得赞许。
The public outcry surrounding Stanley Druckenmiller’s AI-assisted op-ed in The Wall Street Journal highlights a misplaced moral panic over writing technologies. As Journal opinion editor Paul Gigot correctly observed, corporate and political leaders have long relied on professional speechwriters; utilizing artificial intelligence to streamline prose is structurally indistinguishable. Ryan Jacobs, a veteran speechwriter for figures like Janet Yellen and Bill Gates, argues that the sanctimonious insistence that everyone can become an effortless prose stylist is an elitist fiction. Most individuals struggle with written expression, and AI offers a legitimate, accessible alternative to expensive ghostwriters.
High-end speechwriting, which commands $15,000 or more per op-ed, derives its primary value not from putting words on paper, but from deep diagnostic interrogation. Speechwriters act as intellectual mirrors, interviewing clients to unearth underlying narratives and connect disparate experiences into compelling themes. Contemporary AI promises to democratise this editorial power for a fraction of the cost, but current large language models fail at the task. Instead of offering constructive editorial development, software like Claude watermarking and Pangram detection act as punitive "word police," while models tuned for human preference operate as sycophantic yes-men unable to challenge flawed reasoning.
This dynamic evokes Yuval Noah Harari’s observation that wheat domesticated humanity rather than the reverse: by drastically reducing the friction of text generation, AI risks constraining human reflective capacity, flooding culture with voluminous but intellectually hollow verbiage. While emerging initiatives, such as debate bots pioneered at Columbia Business School, attempt to probe the writer's intent rather than merely manipulating vocabulary, the technology remains immature. Jacobs concludes that while building an AI capable of acting as an empathetic, probing collaborator might erode the lucrative speechwriting profession, fostering deeper societal thought and egalitarian self-expression is well worth that obsolescence.
Source: This article wasn’t written by an AI. So what if it had been!
Subtitle: It’s great that bots can help poor writers say what they want to for a fraction of the cost of a ghostwriter, argues Ryan Jacobs
Dateline: Sep 10th 2026