Primary sourceAI Will
View sourceEvent ·Multi-source
Palisade Research 的 Jeff Ladish:别看 40 个基准测试,看 agent 能否真的干完活
Palisade Research 的 Jeff Ladish 认为,衡量 AI 能力的关键不是 40 个基准测试,而是 agent 能否自主完成有经济价值的真实工作。他的转折点发生在 Anthropic 今年 6 月发布的 Fable 模型前后:从偶尔使用 agent 变成每天都在用。他提到 METR 采取让软件工程师执行长周期任务的评估方式,比基准分数更能反映真实能力。他希望回归用真实工作检验 AI 自主完成任务的模式。
Current read
Palisade Research 的 Ladish 说 Fable 让他从偶尔用 agent 变成天天用,他觉得比基准分数更有意义的是 AI 能不能独立把活干完。
11 sources31° AI heatLast updated 2026/11/6 04:17:12
Evidence
11 sourcesRelated source 1Gorden Sun
View sourceRelated source 2IT之家
View sourceRelated source 3lmarena.ai
View sourceRelated source 4Greg Brockman
View sourceRelated source 5Yangyi
View sourceRelated source 6shao__meng
View sourceRelated source 7kimmonismus
View sourceRelated source 8Decoder
View sourceRelated source 9METR
View sourceRelated source 10Gary Marcus
View sourceConflict check
Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.