在一项针对开发者社群的调查中，对AI生成内容的抵触情绪明显浮现。1当检测到文章由AI撰写时，受调查的668名开发者中有78%会立即停止阅读。1这种排斥不仅限于阅读行为——71%的受访者表示将来会避免阅读这些作者的其他作品。1
在内容选择上，读者的倾向更为明确：98%的受访者宁愿阅读作者本人撰写的（即使不完美的）作品，也不愿看经AI润色加工过的版本。1这反映出读者对真实创作与作者真挚表达的重视程度超越了文字打磨的精准度。
技术进步为识别AI写作提供了新途径。1Pangram 4模型在假正率和假负率方面都实现了显著改进。1有观点建议，公开写作平台应当要求内容通过AI检测工具验证为人工撰写，以此维护写者与读者之间的社会契约。1
A significant backlash against artificially generated content is reshaping reader expectations and content policies. Among 668 developers surveyed, 78% reported stopping reading immediately upon detecting AI-generated writing, while 71% indicated they would avoid such authors in the future 1. The preference for authentic work is overwhelming: 98% of respondents said they would choose imperfect, human-written pieces over AI-polished versions from the same author 1.
This sentiment has prompted technological responses to address the problem. The Pangram 4 detection model represents a notable advance in identifying machine-generated text, showing marked improvements in both false positive and false negative rates 1. Building on this capability, some organizations are now implementing stricter content policies. The Oxide team has adopted a requirement that all public writing must pass Pangram detection as human-authored, establishing a verification standard for trustworthiness 1. The underlying concern centers on preserving the fundamental social contract between writers and readers, which AI-generated content threatens to undermine 1.
评论
还没有评论，欢迎留下第一条。