以下部分引用 AI 总结，现阶段 AI 仍然有幻觉。请以内容的原文为准。

数据新鲜度群聊 07-14 ✓卖方 07-14 ✓Wrap 断档15天 ✕News 07-15 ✓X 断档12天 ✕页面生成 07-15

METR_Evals

6 atoms · 跨 6 天 · 首见 2026-04-17 · 最近 2026-07-02

三色: 🟦 fact 6 · 🟥 take 0 · stance ▲0/▼0/◆2

来源: X 6

时态: fresh:6

标签: 好信源:5 · 好数字:2 · 好思考:1

2026-05-11 ⭐ · technical workers self-report AI impact on work value
technical workers claim that today's AI impacts value of their work to an extraordinary degree (& growing over time)
tags: 好信源·好数字 · → daily
value: qty=extraordinary degree · date=2026-05-11
2026-06-10 · 研究表明模型时间视野每几个月翻一番
_@METR_Evals showed that models' time horizons have doubled every few months._
tags: 好信源·好思考 · → daily
value: date=2026-06-10
2026-07-02 · 执行监控能力建设任务
_I joined @METR_Evals to help build out their monitorability effort. Our mission is to measure and inform the world about AI covert capability._
tags: 好信源 · → daily
2026-06-08 · found SWEBench results more than half unmergeable slop
_@METR_Evals found that more than half of SWEBench results is unmergeable slop._
tags: 好信源 · → daily
2026-06-04 · eval duration cap
_METR_Evals cap out at ~16 hours_
tags: 好数字 · → daily
value: qty=~16 hours
2026-04-17 · makes the Most Important Chart in the World
the 30-person AI nonprofit that makes the Most Important Chart in the World
tags: 好信源 · → daily

2026-07-02 · fact · 执行监控能力建设任务 · → daily
2026-06-10 · fact · 研究表明模型时间视野每几个月翻一番 · → daily
2026-06-08 · fact · found SWEBench results more than half unmergeable slop · → daily
2026-06-04 · fact · eval duration cap · → daily
2026-05-11 · fact · technical workers self-report AI impact on work value · → daily
2026-04-17 · fact · makes the Most Important Chart in the World · → daily