如果你造出这么愚蠢的“工具”,为什么要宣传这件事?
If you'd built a "tool" that stupid, why would you advertise the fact?

原始链接: https://svpow.com/2025/10/13/if-youd-built-a-tool-that-stupid-why-would-you-advertise-the-fact/

古生物学家迈克尔·韦德尔收到了一封来自academia.edu的奇怪邮件,称他的关于蜥脚类恐龙神经棘的科学论文被他们的AI转变成了“类比”。该AI将棘的 bifurcation(分叉)比作河口三角洲——韦德尔认为这种比较毫无意义,相当于将研究比作《追忆似水年华》。 韦德尔对AI的滥用感到沮丧,强调了真正有帮助的应用(如ChatGPT提供的编程协助)与这种无意义的输出之间的差距。他质疑为什么平台会*宣传*这种有缺陷的“工具”,并拒绝付费升级以查看更多例子。这一事件凸显了人们对AI生成内容质量和相关性的日益担忧,尤其是在应用于科学研究等专业领域时。

黑客新闻 新 | 过去 | 评论 | 提问 | 展示 | 招聘 | 提交 登录 如果你构建了一个如此愚蠢的“工具”,为什么要宣传这件事? (svpow.com) 11 分,由 surprisetalk 发表于 2 小时前 | 隐藏 | 过去 | 收藏 | 讨论 考虑申请YC冬季2026批次!申请截止日期为11月10日 指南 | 常见问题 | 列表 | API | 安全 | 法律 | 申请YC | 联系方式 搜索:
相关文章

原文

October 13, 2025

A few days ago I got a sensationally stupid email from one of those websites that most of us probably have a subscription to, but which I will not give the oxygen of publicity by linking to[1].

The subject line was:

Your paper “NEURAL SPINE BIFURCATION…” is now an analogy.

No; no, it’s not.

Our paper Neural spine bifurcation in sauropod dinosaurs of the Morrison Formation: ontogenetic and phylogenetic implications (Wedel and Taylor 2013) is not an analogy. It’s a scientific study of whether neural spines bifurcate progressively through ontogeny. (Spoiler: not really. Somehow we stretched that out to 34 pages.)

The email then goes on to say:

Our AI turned your research into an easy to understand analogy

Oh, did it now? Do tell.

The bifurcation of neural spines in sauropods can be likened to the branching patterns of river deltas. Just as a river flows into a delta…

But it really, really can’t. You might just as well say “The bifurcation of neural spines in sauropods can be likened to Marcel Proust’s seven-volume masterwork À la Recherche du Temps Perdu.” It would be exactly as meaningful.

To add insult to injury, they want me to pay to upgrade to Premium if I want to see how this nonsense continues. Hard pass.

When I forwarded this exercise in idiocy to Matt, he replied with just two lines:

What is this I can’t even.

If you’d built a “tool” that stupid, why would you advertise the fact?

Well, quite.

The current generation of “AI”s do have some uses. I ask ChatGPT questions about the minutiae of programming all the time, like when I wanted to know whether there is a standard pattern for invoking a React hook on behalf of a class-based component. But for every useful application of LLMs, there are ten useless or actively destructive uses. I’m not sure which of those two categories “Your paper is now an analogy” falls into.

Just stop it.

[1] It was academia.edu, and who can possibly explain how they were able to get a domain name in the .edu TLD?

 


doi:10.59350/5f44y-g3k32

联系我们 contact @ memedata.com