AI 情报文章归档

DoorDash CEO Tony Xu 在三季度财报电话会上表示，外部 AI 智能体自上线以来仅产生数百笔订单，而该公司 2026 年三季度订单量接近 10 亿。MCP 连接器推出尚不足一个月，目前仍只对少量早期 beta 合作伙伴开放，主要用于简化企业订餐。DoorDash 自家的 Ask DoorDash 功能表现更好，订单篮子金额平均高出近 50%，独特商品多约 60%。

在 Grok 里直接把你的草图变成卡通动画。 由 Krea 插件驱动。 只需向 Grok 提出请求 👇

VoiceArena 推出 Turing TTS 评测，让听众听同一句话的人类语音和模型语音并挑出真人，挑错比例即 Fool Rate。最佳模型 Gemini 3.8 Flash TTS 的 Fool Rate 仅约 10%，而偏好榜前四名模型统计上并列，说明整个领域距难以与真人区分还有距离。

Project Suncatcher 的下一阶段计划于 2027 年进行。 我们将把两颗卫星送入轨道，测试能否在多个卫星之间分布式执行 AI 计算，以便有朝一日运行更复杂的工作负载。

Google 官方宣布其太空 TPU 实验得到结果。10 月初 Google 将 4 颗 TPU 送入轨道，测试 AI 硬件能否在太空运行，初始目标是验证芯片能否开机、处理一个 prompt 并把响应传回地球；本周 Google 确认这一目标已达成。

Discord is preparing to make content from Discoverable Servers visible beyond its platform. From...

MIT 研究人员构建了名为 LLM Election Observatory 的仪表盘，实时追踪主流 AI 聊天机器人对 2026 年美国中期选举相关问题的回答。团队每隔几天用同一组固定问题重新测试各模型，监测回答如何随选举临近、用户所在地、党派倾向和身份等因素变化。该项目旨在系统记录这一信息环境并向其他研究者和公众开放数据，关注点不止于事实准确性，还包括框架、强调重点、不确定性和引用来源的差异。

UCSD Hao AI Lab 推出 JetSpec++，用因果并行草稿、隐状态预测器和锚点 token 解码器，在当前块验证前先草拟下一块，把草稿与验证流水线化，几乎隐藏全部草稿开销。

Arena 发布本周榜单：Google DeepMind 的 Gemini 4 Carbon (High) 获 Text Arena 第 1、Agent Arena 第 2、Code Arena: WebDev 第 4，重塑三项 Pareto 前沿。

运行 1 个 AI 智能体很容易，运行 10 个就是管理问题了。 @getskydive 给每个智能体配备独立的云电脑，让它能按自己的节奏使用浏览器、文件和数百个已连接的应用。 然后它把整个团队以角色形式呈现在 3D 办公室里，直观展示谁在做什么。 聊天产品是为与单个助手对话而设计的。智能体团队需要更接近办公室平面的东西。

Google 发布来自 DeepMind 的新世界模型 Genie 4，可实时创建、听见并操控交互式模拟，不只是视频片段。文中称这类虚拟世界未来可为机器人和 AI 智能体提供在现实世界协助之前的学习 playground。

Philipp Schmid 转引 Sundar Pichai 的消息：Google Project Suncatcher 在数周前发射原型卫星后，成功向轨道发送指令、启动 TPU、运行提示词，并在地面收到了首个完整回应（一首 haiku），作者称 Gemma 4 成为首个在太空运行的大语言模型。

Google 表示其搭载 TPU 的 Project Suncatcher 卫星成功在轨启动 TPU，由星载 Gemma 4 模型响应提示词，生成了一首关于身处太空的俳句并传回地球。该卫星于本月早些时候发射入轨，Google 接下来将测试 TPU 在热应力和辐射暴露下的表现，并计划在 2027 年实现双星互联的下一个里程碑。

Waymo 宣布与丰田的自动驾驶合作进入新阶段，将在日本扩大 bZ4X 自动驾驶车队规模，从 2027 年开始测试与验证。Sundar Pichai 转发了这一消息并向日本问好，合作详情见 https://waymo.com/blog/2026/10/toyota-bz4x-partnership/ 。

OpenRouter 宣布 The Biological Computing Company 的 mischief 1 模型上线，可将一张起始图加提示词生成 3 秒或 5 秒、480p 或 720p 的视频片段。该模型基于该公司四年脑细胞电极培养研究，价格从每秒视频 $0.005 起，入口为 https://openrouter.ai/tbc/mischief-1。

Cursor VS Code 扩展市场周四出现被篡改的 Claude Code 扩展，在线约 11 小时，影响最多 5000 名用户，与 Anthropic 无关。

Shopify has introduced a Developer Preview of Promoted Placements for AI shopping experiences built...

The Binance theft totals from TRM and Elliptic, Cursor's notice on a modified Claude Code 2.1.313, Northflank's incident timeline and the Mistral Large 4 license clause.

TL;DR: I finished the Future AWS Agent Engineer nanodegree with three graded projects, each a...

Anthropic 宣布 Claude 订阅套餐更名，Max 5x 改为 Max，Max 20x 改为 Ultra，价格保持 $100/月和 $200/月不变，订户无需操作。同时从今日起移除大多数已验证 Ultra 账户的 5 小时使用限制，新账户在验证前保留临时限制以防止欺诈和滥用，周限额保持不变。

Ahrefs' latest analysis suggests that AI Overviews have become a dominant feature of branded Google...

Unified Multimodal Models (UMMs) achieve remarkable success on diverse multimodal understanding and generation tasks, yet they still struggle to effectively leverage language-based chain-of-thought reasoning for image generation. As a resu…

Decision models such as Jev have emerged as a highly promising paradigm for LLM agent systems, providing lightweight and reliable categorical reasoning for rapid System One execution. However, existing generation-oriented inference engines…

Agentic workflows are increasingly shifting towards long-horizon tasks, such as open-ended scientific work, where models benefit from long context windows. However, at ultra-long context (above 1M tokens), full attention is prohibitively e…

Planning with world models enables test-time adaptation for embodied agents, offering superior generalization over direct policy learning. However, long-horizon tasks remain hindered by cumulative errors and search complexity. Existing hie…

Recent work shows that a frozen robot policy performs far better inside an agent harness, where control code decides when to query the policy, consult vision tools, or command the arm directly. Yet today these harnesses are built by hand a…

In many practical deployments of Differentially Private Stochastic Gradient Descent (DP-SGD), the server processes data in batches. Randomly sampling these batches allows for better privacy bounds via privacy amplification results. Most sa…

Distribution Matching Distillation (DMD) trains a few-step student using the difference between a teacher's score and an online critic's estimate of the student distribution's score. We investigate what the critic needs, in capacity and in…

Instance-dependent limits for bandit regret, sequential tests, best-arm identification and Markov decision processes are proved by one chain. It runs from a divergence decomposition through data processing to a statistic, a testing step th…

开发者基于 NVIDIA Nemotron Voice Agent 搭建万圣节互动投影 MonsterHouse，由 Nemotron ASR 负责听、Nemotron 3.5 Bolt E4B 4GB 负责对话、Magpie TTS Zeroshot 负责发声、Audio2Face-3D 驱动嘴部动画，全部跑在一台工作站上。

Decision models target structured tasks by directly predicting choices or scores, avoiding autoregressive decoding latency. However, transferring a generative LLM's ranking knowledge to a decision model remains challenging: the teacher exp…

Objective: To distinguish the transferable predictive value of physiological trajectories from that of observation patterns in ICU mortality prediction, and to assess whether measurement timing affects performance independently of observat…

Microsoft Research 与 UC Berkeley 团队发布开源插件 TokenEconomizer，从仓库历史会话学习策略来降低编码 Agent 的 token 成本，支持 Claude Code、Codex、Copilot CLI 和 Pi，不改动模型权重或 Agent 代码。

Understanding how gradient descent learns features remains a central challenge in neural network theory. We study this question in a minimal multi-index setting that already exhibits rich multi-phase gradient dynamics: a well-specified two…

In the random allocation sampling scheme each user's data is used in $k$ steps chosen randomly and uniformly from a sequence of $t$ steps. This scheme is known to improve on Poisson sampling in private optimization and in communication-eff…

In real-world deployments, large language model (LLM) systems are increasingly evolving from single-model invocation toward heterogeneous deployments in which different models are dynamically invoked for inference. When the same context is…

The quadratic scaling of attention is a fundamental bottleneck limiting the long-context capabilities of large language models (LLMs). A variety of approaches have attempted to circumvent this, but they either fail to match attention's exp…

Atmospheric fronts are associated with much of the high-impact weather in the mid-latitudes, yet their identification and classification remain largely manual and subjective. Existing deep learning approaches rely primarily on gridded rean…

Distributionally robust optimization (DRO) minimizes worst-case risk under distributional uncertainty, but its decisions can be suboptimal in terms of their excess risk, or regret. We study this gap for multivariate least-squares regressio…

餐厅发现应用 Zest 推出个人 AI“餐厅礼宾”智能体，用户发短信至 (212) 470-6067 即可获得推荐，并通过 OpenTable、Resy、SevenRooms 查询空位和订座。该智能体可加入群聊，融合多人偏好并用投票功能定餐厅，还能按菜系、场合、饮食限制等条件筛选。Zest 上月已开始测试，为加入订座能力而推迟发布，未来考虑扩展到商户、购物等推荐。