奇文共赏 - THE 2028 GLOBAL INTELLIGENCE CRISIS

这有点抽象了,不太现实。IP只是一个例子,随便有个小问题就绕不过去。反正我没看到AI发展了这么几年时间里爬虫有什么进步,还是无头浏览器搞一搞,不成功就上selenium模拟点击继续搞。反而cloudflare之类的CDN的反爬虫技术越来越好了。

他们最近在推专门的反AI爬虫功能
https://www.cloudflare.com/zh-cn/learning/ai/how-to-block-ai-crawlers/

To do this, we decided to use a new offensive tool in the bot creator’s toolset that we haven’t really seen used defensively: AI-generated content. When we detect unauthorized crawling, rather than blocking the request, we will link to a series of AI-generated pages that are convincing enough to entice a crawler to traverse them.

爬虫 国内好像是违法行为。。。AI 犯罪人还是要吃官司的吧

瞎爬 很容易 就被限流了

Consumer agents began to change how nearly all consumer transactions worked.

Humans don’t really have the time to price-match across five competing platforms before buying a box of protein bars. Machines do.

Travel booking platforms were an early casualty, because they were the simplest. By Q4 2026, our agents could assemble a complete itinerary (flights, hotels, ground transport, loyalty optimization, budget constraints, refunds) faster and cheaper than any platform.

是不是过几年也没有信用卡毛可以薅了

substack多弄几个也太贵了

泥潭有人愿意每个月付50吗

1 个赞

$900包年吧,之前有活动来着

草,比我买burry还贵

The Information, SemiAnalysis, Citrini Research, The Pragmatic Engineer, 都时不时会有点干货

1 个赞

应该用泥潭数据训练羊毛agent 效果杠杠的 :troll:

这几个我都关注但是free

Substack 系统太烂了,每月50不贵,只是读不过来,除了几个特别关注的我现在都用ai summary。

1 个赞

他那个系统没办法统manage subscribe 太傻逼了

真的有点扯、现在c1s其实就能做到这个了

这篇写得可好多了,至少薅羊毛那段和泥潭息息相关

第一章

FinTech的裂痕aka 风控系统的系统性失效

历史的转折点往往没有震耳欲聋的轰鸣,它只是一页再也无人维护的风控规则,以及不知名的Telegram群聊里那句轻描淡写的:“不是哥们,Square的风控居然都被裁了,速薅!!!”

A scenario in which productivity surges but aggregate demand collapses while measured output rises violates accounting identities

看到这句话,跟Gemini分析了一下

只能认为这篇文章在放屁

按照它的逻辑1929从“会计”规律上就不可能发生

全要素生产率在1918-1929暴涨,但是measured output到股市爆炸前才开始下行的

也许我们可以反过来思考:agent不能爬到的网站我就懒得去定了反正价格一般不会差到离谱大,那那些愿意开放爬虫甚至接入api的网站就赢了。

不要低估人类的懒惰

趋势应该是是每个平台会做自己的Agent,帮你一键定。

很多OTA都做了,比如TripAdvisor