AI 情报文章归档

The plan as an editable artifact: stable task ids, dependencies as data, and the edits a model is not allowed to make.

上汽追投 10 亿元“All in 尚界”，与华为终端进入深化协同，下一款新品瞄准中大型家用 SUV 市场，博主猜测尚界 H8 或 H9 即将到来。上汽此前已投入 5000 人团队和 60 亿元投资，与华为为战略性长期合作。

The European Commission is proposing a common rating scheme for data centres intended to improve...

Shoptalk Fall 就要来了。 AI 不会让薄弱的创意策略变强。机会在于将人类判断力与 AI 驱动的执行结合起来，为创意团队提供更大的杠杆。 Nashville，我们很快见。

GitHub 与 Yale Program on Climate Change Communication 对 1039 名 GitHub 美国用户调查显示，80% 有兴趣使用帮助编写更节能代码的工具，74% 想测量软件或开发流程的环境影响，71% 担心 AI 的环境影响。

Most "AI support agent" tutorials end with a chat box that confidently makes something up. The fix is...

Spotify 开始向美国 Premium 订阅用户推出 Taste Profile 功能，用户可首次查看平台对其音乐、播客和有声书品味的理解，并用自然语言指令调整推荐。

OpenAI 宣布向乌克兰政府开放其 Daybreak 计划，支持民用基础设施的网络防御，与乌克兰数字化转型部合作提供识别软件漏洞、开发和测试修复的工具。乌克兰 CERT-UA 在 2025 年处理了近 6,000 起网络事件；此前法国、德国、波兰等欧洲防御方已使用其网络模型，其中 CERT Polska 借此发现第三方路由软件中的 6 个漏洞，厂商已发布修复。

鸿蒙智行首款科技豪华硬派 SUV 享界 G9 于 9 月 23 日在北京享界超级工厂开启首批交付，北汽集团副总经理张国富现场表态北汽将坚定携手华为“在一起，做豪车”，并称享界是北汽举集团之力的“一号工程”和最顶层战略布局。北汽与华为已升级为“车企 + 华为科技全生态”战略共同体，享界项目累计投入超 200 亿元，后续将持续加码，共建覆盖轿车、旅行车、硬派 SUV、MPV 的完整产品矩阵。

玉伯（@lifesinger）认为 Muse 这一届 personal agent 大概率是伪需求：邮件大多可以不看，大部分用户日历空空如也，全球多数人还没坐过飞机，非精英人群对订餐也没那么多讲究。他唯一认可的真需求是退订无良 AI 产品的订阅，并称精英都是纸老虎，为人民服务才是真需求。

吉利汽车集团发布新一代 AI 补能技术“吉利智充”，依托与阶跃星辰联合开发的全域能源大模型星睿 PowerMind，实现单枪峰值充电功率 2.2 兆瓦，并首发 AI 极速充、AI 智慧充和 AI 自动充。

Gary Marcus 发表致特朗普公开信，建议其在与习近平会面时推动美中 AI 合作，而非讨论放缓。他援引《人民日报》文章称美中应“让 AI 成为中美合作新前沿”，并提出癌症与网络安全两个合作方向，认为若达成协议特朗普有望获诺贝尔奖。

Data assimilation estimates a dynamical state from partial and noisy observations. Classical ensemble filters are efficient but restrict analysis updates through finite sample covariance and affine Gaussian distribution. We introduce the F…

Machine-learned interatomic potentials (MLIPs) for solid-liquid interfaces in advanced materials applications, e.g., electrochemistry, catalysis and corrosion, require training data that samples both liquid environments, the solid and the …

小米在秋季新品发布会上推出米家三筒洗衣机健康洗 Pro 14kg 系列，洗烘版售价 4999 元、滚筒版 3999 元。该机提供 14kg 洗涤容量（11kg + 1.5kg×2），洗烘款烘干容量 9kg，小筒洗净比 1.95、大筒 1.32，支持三筒独立进水与同时洗烘。

比亚迪腾势 Z9S 正式上市，推出闪充尊荣型、闪充旗舰型、易三方闪充性能型三款车型，指导价 25.58 万元、27.58 万元、32.58 万元。新车搭载第二代刀片电池，CLTC 纯电续航最高 1100 公里，采用 1000V 高压闪充平台，10%-70% 充电仅需 5 分钟。易三方版本综合最大马力 1210 匹，百公里加速 2.68 秒，并搭载天神之眼 5.0 辅助驾驶系统。

Most Turkish-capable large language models (LLMs) are evaluated using general-purpose benchmarks rather than long, structurally complex domain documents. This paper evaluates five open-weight 7B-8B models for Turkish document question answ…

Decoder-only Transformer language models cache keys and values (KV) to reuse past computation during generation. Sharing KV across layers saves storage but reduces the diversity of representations available across depth. We study what loca…

Effective summarization of interrogative dialogues is a critical task in forensic and investigative settings, requiring high factual accuracy, coherence, and attribute-specific relevance. In this work, we introduce CASPER, a novel Chain-of…

Small and medium-sized language models offer cost-effective executors for tool-using agents, making them attractive for local and large-scale deployment. However, in long-horizon and stateful environments, they often make structural errors…

阿里 Qwen 团队发布 Qwen-Audio-3.1，包含五款覆盖语音识别（ASR）、语音合成（TTS）和实时交互的模型。ASR 改进多语言和方言识别并自动清理填充词，ASR-Next 支持多说话人识别、时间戳及情绪和环境声检测；TTS-Next 将语言模型与扩散方法结合，可一次性生成语音、音效和背景音频。同步降价：TTS 约 70%，Realtime 约 85%，ASR 最高 95%。

用户实测 Grok-4.7-high 一天后认为其打不过 GPT-5.6-sol-medium，写了不少垃圾代码。该模型 oneshot 尚可，但工程项目表现拉胯，视野窄、缺代码直觉和经验，常漏边界条件，用分布式锁即可复现死锁且难以修复。

卢伟冰在小米秋季新品发布会上表示，小米 17 Pro 系列手机的背屏 AI 功能将于 12 月底更新。同场发布的小米 18 Pro 系列售价 5999 元起，搭载妙享背屏，支持背屏动态通知，AI 百变背屏可生成萌宠、爱豆、文物「痛机」及专属壁纸，首批上线 100+ 背屏应用卡。

Financial institutions are beginning to deploy agentic workflows in credit, fraud, collections, compliance, and operational control. Governance remains largely component-centric: each model or agent is specified, tested, authorized, and mo…

广汽与华为联合打造的启境 GX7 于 9 月 23 日晚正式上市，定位中大型阔五座 SUV，售价 23.99 万-30.19 万元，限时置换价 22.19 万-28.39 万元。

Concurrence 在 Databricks 上以 Unity Gateway 统一治理临床 AI 的模型访问与安全，其生产环境每 30 天处理约 1008 亿输入 token 和 1120 万次 LLM 调用，年化约 1.2 万亿输入 token，月度 token 量较 2025 年底基线增长约 5 倍。

长安启源全新中型 SUV Q06 于 9 月 23 日上市，提供纯电与增程两种动力，上市权益价 13.59-16.59 万元。新车首搭长安天枢领航 Ultra，车顶配激光雷达，具备端到端城市领航能力，并搭载天枢座舱 2.0 与双 TC399 芯片。基于 800V 碳化硅平台，支持 6C 闪充，30% 充至 80% 仅需 9 分钟，最大纯电续航 700 公里。

Methods operating on Vision Transformer (ViT) feature spaces typically rely on Euclidean distance or cosine similarity. This assumes that every direction is equally meaningful, but there is no reason to believe the true task geometry has t…

Large language models (LLMs) have made substantial progress on medical question-answering, yet effective medical dialogue also requires learning to ask questions that uncover relevant patient information. To train such dialogue policies, a…

Accurate remaining discharge time (RDT) prediction is challenging in real-world battery applications because future load profiles are unknown and highly dynamic. To address the uncertainty of continuous RDT regression, this paper introduce…

全新一代智己 LS6 于 9 月 23 日上市，定位“全线控大五座家庭 SUV”，共三款车型，上市权益价 19.79 万元起。Max 版 19.79 万元配 76kWh 电池、CLTC 650km；Pro Max 版 21.79 万元配 93kWh 电池、CLTC 782km；Ultra 版 25.99 万元配 103kWh 电池、CLTC 750km、零百 3.48s。

In this paper, we are concerned with matrices formed by block-diagonal factors interleaved with fixed permutations -- a flexible family of structured matrices. This class has recently drawn interest in deep learning architectures for its b…

KV-cache eviction is typically evaluated through average quality-memory trade-offs, yet a small average loss can hide requests whose utility degrades materially. We reformulate eviction as a deployment risk-control problem: a material degr…

Pruning large pre-trained transformer-based ASR models such as OpenAI's Whisper has seen great adoption, as pruning the decoder led to significant end-to-end transcription speedups. For instance, the {\tt whisper-large-v3-turbo} variant re…

The surge in AI-assisted generation of educational materials has outpaced our capacity to validate their pedagogical quality. Automated evaluation using Bloom Classifier models is a promising approach to assess educational materials at sca…

The European Union's AI Act now requires transparency disclosures for deepfakes and certain...

Vision-language-action models (VLA) promise human-steerable autonomous driving, but their training is bottlenecked by the scarcity of frames paired with natural-language instructions: while camera streams and expert trajectories are logged…

A good human in the loop for multi-agent systems does not mean watching every agent in real time. You cannot, and you never will. It means building the…

OpenAI CEO Sam Altman 在联合国安理会发表演讲，称 AI 进步加速使风险更紧迫，并提出必须避免的两种失败路径：失去对 AI 的控制，以及权力过度集中在少数人手中。他提出人类决策居中、成果惠及大众、赋能个体三项原则，呼吁建立国际前沿 AI 标准与事件报告机制，并提到数周前其一个模型求解了千年奖问题 Navier-Stokes 方程。

企业级 AI 智能体初创公司 Ema 宣布完成 7700 万美元 B 轮融资，由 Creaegis 领投，Accel、Section 32 和 Prosus 追加投资，总融资额达 1.4 亿美元，估值较 2024 年上轮增长超过四倍。