一位顾问发现，尽管他做了基础工作，但使用 AI 撰写客户提案导致客户认为其是敷衍和不真诚的。这种经历表明，读者可以通过微妙的线索（如不自然的流畅度和结构）来检测 AI 生成的内容，这通常比任何 AI 检测工具的标记都要早。作者认为，专注于提示词以使 AI 听起来像人类，不如确保 AI 的输出反映真实的人类视角和努力有效，并将当前的趋势比作横幅广告盲视的早期阶段。 AI
影响 读者正在发展“AI 盲视”，类似于广告盲视，导致他们不顾检测工具而忽视 AI 生成的内容。
排序理由 由特定个人撰写的观点文章，讨论了 AI 生成内容的感知质量和可检测性。
在 Mastodon — fosstodon.org 阅读 →
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
How I Multiply Myself Without Shipping Slop: A Solo Founder’s AI Content Workflow The content creation workflow I built so I could explain Siren without burning out. Voice profile, interview, multi-pass editing, customer avatar agents. No slop. https://www. alexstandiford.com/art…
AI Doesn’t Have Perspective (And Your Readers Can Tell Before Any Detector Can) Detector tools didn't catch my AI proposal. My client did. Here's what your readers notice that the detectors miss, and how to fix it without bypass tricks. https://www. alexstandiford.com/articles/es…
AI Blindness Is the New Ad Blindness (And No, You Can’t Prompt Your Way Out) Most 'how to make ChatGPT sound human' advice is a magic prompt. It doesn't work. Here's why your readers can spot AI from across the room, and what actually fixes it. https://www. alexstandiford.com/art…