AI Futures Project 更新时间线：超级人类 AI 研究员或于 2027 年出现

UC Berkeley 这篇论文说清了一件事：降 agent 成本不一定非换模型，改 harness 层就能差出 2.6 倍。他们还开源了 Agency 这层中间件，写一次优化就能套在 Claude Code、Codex 上。跑 agent 产品、被 API 账单压着的团队都该看看，先删重复上下文再说。

DAIR.AI@dair_aiSource

让编程智能体去加速 LLM 训练很容易做出假提速，这篇论文的做法是把 workload、loss、计时先用契约锁死再动手改代码。MFU 从 17.4% 提到 30.0%，最高 2.75 倍加速。做训练 infra 或跑大模型集群的工程师很值得细看。

DAIR.AI@dair_aiSource

The non-obvious finding is that distractor damage is localised in a handful of attention heads, so a lightweight patch can match most of what a costly retraining effort would achieve. Teams running RAG or agent systems on open-weight models can implement the mask as an inference-time intervention without touching training pipelines. Compared with context-cleaning heuristics that filter noisy documents before inference, head masking operates inside the model and needs only a one-time identification of the offending heads per model. Practitioners in ASEAN serving models under cost constraints benefit because the fix adds no training expense.

rohanpaul_ai@rohanpaul_aiSource