2026-09-11 科技日报

扫描 592 篇候选内容 · 覆盖 650 个信息源 · 纳入 576 篇

📌 今日要点

• Agent Evaluation Metric for multi-turn conversations
• Jensen Huang explains why Nvidia will grow an astounding 70% next year
• Mark Wahlberg is coming to TechCrunch Disrupt 2026, and he wants to talk about your work, not his
• OpenAI puts Pro subscriptions on hold due to Astra demand
• Anthropic details distillation campaigns from Alibaba, Moonshot AI, and DeepSeek
• Meta’s AI agent Muse is now the No. 2 app in the US

📝 科技简讯

• Agent Evaluation Metric for multi-turn conversations — AWS Machine Learning
• Jensen Huang explains why Nvidia will grow an astounding 70% next year — TechCrunch AI
• Mark Wahlberg is coming to TechCrunch Disrupt 2026, and he wants to talk about your work, not his — TechCrunch AI
• OpenAI puts Pro subscriptions on hold due to Astra demand — TechCrunch AI
• Anthropic details distillation campaigns from Alibaba, Moonshot AI, and DeepSeek — TechCrunch AI
• Meta’s AI agent Muse is now the No. 2 app in the US — TechCrunch AI
• Anthropic reveals rogue AI agents hate CAPTCHAs, just like you — TechCrunch AI
• India’s Pocket FM doubles revenue run rate to $500M as AI powers 93% of audio content — TechCrunch AI
• How a researcher uses Codex and ChatGPT to search for new antimicrobial molecules — OpenAI Blog
• Now everyone can put data to work — OpenAI Blog
• Introducing ChatGPT for Financial Services — OpenAI Blog
• Expanding AI access and cyber defense for federal, state, local, and tribal governments — OpenAI Blog
• Introducing the Agents API — OpenAI Blog
• Build more natural voice experiences with GPT‑Live‑1 in the API — OpenAI Blog
• AhaBench: Do Agents Learn from Prior Experience? A Benchmark for Long-Horizon Continual Learning — ArXiv ML (cs.LG)
• When Do Options Help? Policy Necrosis and Redundant Coverage in Option-Critic — ArXiv ML (cs.LG)
• Multi-granularity Adaptive Hypergraph Representation Learning via Granular-ball — ArXiv ML (cs.LG)
• Capsule Lens: Locating and Tracking Concept Geometry in Model Representations — ArXiv ML (cs.LG)
• HB-PVI: A Hierarchical Bayesian Personalization and Value-of-Information Framework for Complex Activity Recognition — ArXiv ML (cs.LG)
• Endogenous Exploration in Reinforcement Learning with Intrinsic Curiosity — ArXiv ML (cs.LG)
• Robustness of LLM-Generated SystemVerilog Assertions to Semantics-Preserving RTL Transformations — ArXiv ML (cs.LG)
• How to 5x Your Communication Effectiveness with Claude Code — Towards Data Science
• What SHAP Can’t Explain About Agentic AI Fraud — Towards Data Science
• Optimizing LLM Inference Costs in Multi-Agent Systems with Adaptive Model Routing — Towards Data Science
• Who Questions What Works: When Should We Retest Our Assumptions? — Towards Data Science
• ECCV 上，顶尖学者们开始研究如何让 AI 做生意了 — 量子位
• 全球首个 3D 原生城市世界模型 ABot-Earth 0.7 发布，构建 AI 理解真实世界的入口 — 量子位
• 全球首个可仿真的人–场景交互重建框架 HSImul3R：让人类视频真正成为机器人技能来源 — 量子位
• 这个新开源的世界模型只有 1.3B，单卡就能实时跑！ — 量子位
• AGI 时代的第一个生图模型，ChatGPT Images 2.5 上线 — 量子位
• 营销科技巨头蓝色光标与全球达人营销 AI 平台 AhaCreator 达成深度合作，让品牌更高效连接全球 500 万创作者 — 量子位
• 一周连发 6 个模型！这家公司把具身智能的闭环跑通了 — 量子位
• 打造 10 万卡国产算力集群推出 JoyAI 世界模型，京东发布物理 AI 建设最新成果 — 量子位
• 实测星火 X2.5：手搓粒子月亮、拆完 61 页财报……还顺手揪出了我的 Bug — 量子位
• OpenAI’s GPT-Live-1 API lets developers build apps that talk and listen at the same time — The Decoder
• Swarmchasers hunt rogue agents, Anthropic investigates itself, and the trail they both follow is going dark — The Decoder
• Former Deepmind PR staffer says the lab once banned public discussion of AI extinction risk — The Decoder
• Claude Fable 5.1’s language is less “load-bearing” than its predecessor’s — The Decoder
• GPT-6 Astra gives mathematicians a breather, and OpenAI says that’s by design — The Decoder
• New Deepseek model V4.1-Flash cuts memory needs for AI agents — The Decoder

📊 数据概览