三条 7 月 30 日至 8 月 1 日凌晨的公开观点,从家庭日程、组织吞吐到模型外围工程,观察 AI 能力真正进入工作与生活时,哪些系统必须一起改变。
01|把日历变成播客
connect your family calendars and explain your kids' interests.every morning for the drive to school, have it make a podcast that talks about one kid's soccer game that afternoon, one kid's upcoming birthday, some news, etc.—— Sam Altman,2026 年 8 月 1 日 00:01,X 原帖
Sam Altman 分享了一个 ChatGPT Work 的家庭使用场景:把家庭日历和孩子的兴趣接入系统,再在每天上学路上生成一档贴合当天安排的播客。重点不是把信息放进聊天框,而是让 AI 先替人整理「今天值得听什么」。1
Loading content card…
02|组织会被产出撑住
A real issue with AI adoption is that organizations are built around a narrow expected range of human productivity in a role.Too little output is a problem, yes, but way too much can sometimes be as bad since approvals, staffing models & coordination systems cannot absorb it.—— Ethan Mollick,2026 年 7 月 31 日 05:05,X 原帖
Ethan Mollick 把 AI 落地的瓶颈从模型能力移到了组织外围:审批、人员配置和协调系统,可能都按一个狭窄的人类产出区间搭建。模型让一个人产出更多,并不代表旧流程能同步接住这些工作。2
Loading content card…
03|模型之外,还有低垂果实
The low hanging fruit on harness engineering is insane.[...] there's an interesting area of science/engineering on how to best study the effect of harnesses, from post training through eval/inference. Will be a fairly impactful area (in cost savings per performance)—— Nathan Lambert,节选自 2026 年 7 月 30 日 11:18 的 X 原帖
这里的 harness engineering,可理解为围绕模型设计工具、上下文、反馈和运行流程。Nathan Lambert 认为,这些外围条件从后训练到评测、推理都值得被系统研究,性能提升不一定只来自更大的模型。原帖中的
[...] 表示卡片对原文做了节选。3Loading content card…




Comments
Sign in to comment.