[思维分享] AI 创业公司,最有价值不再是数据,而是 gold standard
原文:
(来自YC conference 交流的观点)
One thing I noticed this weekend is that a lot of the talks that the founders gave were around evals and testing, which I don't think that would have ever been true at a previous YC conference.
-
Testing was sort of this afterthought thing that you tried to do as little of as possible.
I heared one really interesting comment from a founder who's building an AI agent who said that he thinks that the most
valuable thing that his company has built is not the code base, it's the eval set.
It's a gold standard labeled set of data of this is the correct answer for the AI to do.
-
And that was sort of a mental change for me that there's ,I think this perception that companies have data assets, but just general
random data is actually not that valuable.
The thing that's really valuable is a gold standard, meticulously(精心标记) labeled eval set.
-
---分割线---
就像现在,任何人都可以调用 AI 大模型,去完成任务,做各种动作,,但最终胜负手 还是在 taste 品味。
eval 就是 taste, 好的,优美的,beautiful,useful
(作为补充,有家设计师 已经完全放弃 用 Figma 做原型,而是依赖 Claude 产生,因为 Claude 有足够的 taste 去生成好的)
我看了看一个AI原画网站,确实牛逼,甚至都能让人物流畅动起来。楼主你在这领域研究蛮多,有从事相关工作或者自己有显卡训练吗?
我去年有搞过AIGC,虽然我有经验但是商业化很难现在
C: 我看了看一个AI原画网站,确实牛逼,甚至都能让人物流畅动起来。楼主你在这领域研究蛮多,有从事相关工作或者自己有显卡训练吗?
我没有做过,只是觉得自己了解还不够,所以多了解点,和 AI 相关我都尽量关注下。
不过我从 23 年初 GPT 火了,就开始自己看 deep learning 的书,可能理解和一般人不太一样,加上有很多关注的领域(如文字,3d打印),其他的地方需要 AI 的协助,因此兴趣 也自然浓一点。
-
至于创业,个人看法,AI 肯定 有用,但未必具备「商业完备性」,
一个例子:比如 最近 Anthropic 公司 最近推出的 Citations 功能(1-24),,我觉得,才让 RAG 这个功能,具备了一些 商业完备性 ,(不但能标注信息来源文件名,还能把 原始的句子也标注出来),这个如果靠自己要实现 是非常难的,效果也未必好。
如果没有 Citation 功能,单单 RAG + AI 已经很让人惊叹了,但确实不够,还没那么好用,用户可能就不愿去用,这也是事实。