Antara Haldar
advocates a radical rethink of development, explains what went right at the recent AI Safety Summit, highlights the economics discipline’s shortcomings, and more.
The prevailing narrative that frames Israel as a colonial power suppressing Palestinians’ struggle for statehood grossly oversimplifies a complicated conflict and inadvertently vindicates the region’s most oppressive regimes. Achieving a durable, lasting peace requires moving beyond such facile analogies.
rejects the facile moralism of those who view the ongoing war through the narrow lens of decolonization.
The far-right populist Geert Wilders’ election victory in the Netherlands reflects the same sentiment that powered Brexit and Donald Trump’s candidacy in 2016. But such outcomes could not happen without the cynicism displayed over the past few decades by traditional conservative parties.
shows what Geert Wilders has in common with other ultra-nationalist politicians, past and present.
剑桥—人工 智能(AI)正在快速发展。人们正在使用生成式AI和大型语言模型(LLM)来构建新服务,执行现有任务,底层技术本身也在迅速发展。正如诺贝尔经济学奖获得者迈克尔·斯宾塞(Michael Spence)所观察到的那样,在经历了近二十年的增长乏力之后,这一波AI部署浪潮很可能会带来显著的生产率提高。每天都有新的消息,比如谷歌最近宣布其AI帮助美国航空公司减少了54%的凝结尾迹,减少了每个航班的气候足迹。
但消息并不全是好消息。就目前情况而言,AI更有可能帮助大型科技公司巩固其主导地位。他们拥有开发和维护最强大的AI模型的资源,并且已经迅速采取行动,将LLM与现有服务捆绑在一起。而与此同时,世界各地的反垄断机构越来越关注科技公司的市场力量。
平心而论,一些评论员——包括一位谷歌工程师在一份内部备忘录中所写——认为这种担忧被夸大了,因为开源LLM的存在在技术上允许任何人在市场上竞争。但即使规模较小的新进入者如雨后春笋般涌现,大型科技公司的主导地位看起来仍然很稳固。最近的一篇论文将开源模型与大型科技公司提供给第三方的AI应用程序编程接口(API)服务进行了比较,发现后者在大多数标准上的表现要好得多。
To continue reading, register now.
Subscribe now for unlimited access to everything PS has to offer.
Subscribe
As a registered user, you can enjoy more PS content every month – for free.
Register
Already have an account? Log in