# Thinking Machines 谈模型自我改进中的人类判断

- 来源：Thinking Machines (@thinkymachines)
- 发布时间：2026-09-12 05:56
- AIHOT 分数：44
- AIHOT 链接：https://aihot.news/items/cmtxiiwic0amyro6p39csztuw
- 原文链接：https://x.com/thinkymachines/status/2098531092286611828

## AI 摘要

Thinking Machines 的 John Schulman 与 Dwarkesh 对谈，讨论模型持续进步与自我改进时人类判断仍不可替代：教模型处理混乱的真实世界任务、用品味判断什么长期有效，以及最关键的——明确说清我们到底想要什么。

## 正文

Our own @johnschulman2 talks with Dwarkesh about where human judgment still matters as models improve and self-improve: teaching them to handle messy real-world tasks, applying taste to what works in the long run, and, above all, specifying what we actually want.

### 引用推文

> Dwarkesh Patel：New episode with @johnschulman2, @oneill_c and @BerenMillidge. I got together with some of the most insightful AI researchers I know who are at the openish comp...