AI 情报文章归档

由前 Anthropic 员工 Rune Kvist 和 METR 前 COO Rajiv Dattani 创办的 AIUC 宣布完成 4000 万美元 A 轮融资，由 Ribbit Capital 领投，总融资达 5500 万美元。

Agility Robotics 发布面向仓库和工厂的新版人形机器人 Digit 5，称其可借助 AI 和传感器识别人并停下或避让，在无安全围栏情况下与人协作。Digit 5 提升负重至 22.7 kg（比前代多 40%），充电 9 分钟可运行 90 分钟；Agility 称订单超 3 亿美元，首批 2027 年初交付，包括欧盟，俄勒冈州 Salem 工厂年产能可达 1 万台。

Thermal optimization of 2D CFET inverters requires testing structural proposals against their electrical costs. We examine these research tasks using an AI agent workflow within a supplied electrothermal model. At 12 nm, Astra selects a re…

在 Anthropic CEO Dario Amodei 发文《We Must Pace the Frontier》呼吁全行业限速后，推特创始人 Jack Dorsey 发布《open the frontier》针锋相对，主张前沿不属于任何大公司。

Large language models can easily generate explanations for their own outputs, but such self-explanations are not necessarily faithful to the model's behavior. We study this issue through counterfactual self-explanations, where a model mini…

有传闻称 GPT-6 Sol 即将发布，与 Tibo 称本周将有多项重大发布、规模堪比 Dev Day 的说法吻合。据其理解，GPT-6 Sol 定位将低于 Astra，类似 Opus 低于 Fable，可能以接近 Astra 的性能提供更好的性价比；GPT-6 Luna 同样值得期待，因 5.6 Luna 在价格与性能上表现极佳。Opus 5.2 也即将到来，更新节奏持续。

Vals AI 让 Claude Fable 5.1 尝试破解苏格兰作家 Thomas Urquhart 1653 年留下的两行数字密码，全程无人提示，耗时 44 分钟、用了 17.6 万 Token 成功解开。

博通 CEO 陈福阳在 CNBC《疯狂金钱》采访中回应 AI 放缓担忧，表示维持长期收入目标不变，算力需求依旧旺盛。此前 Anthropic CEO 阿莫代伊发文主张适度放缓大模型研发，导致博通股价下跌 4.8%。

Vision-language-action (VLA) systems already bring together two valuable resources for robot learning: rich visual representations and demonstrations of successful task execution. Intrinsic Robot Rewarding (IRR) proposes to use these resou…

Modelling with mathematical formalisms like logical formulas, mathematical equations, or regular expressions is an important yet challenging task for students of computer science and other STEM disciplines. Identifying common mistakes occu…

A common small-model deployment runs one shared backbone with several LoRA specialists that answer over the same context. Serving them naively re-prefills that shared context once per specialist. We study a narrow, practical question: for …

Generative AI promises natural language access to the massive numerical telemetry of data centers and Industry 4.0 installations, yet text-to-query and tool-using agents stay unreliable: even frontier models answer little more than half of…

《The Information》援引知情人士消息称，字节跳动今年上半年净利润降至 200 亿美元，降幅为个位数百分比，主因是 AI 投入增加；同期营收 1，200 亿美元，同比增长约 30%。公司还扩张 AI 云业务，近期从银行借入约 300 亿美元，创历来最大单笔贷款，或用于支持 AI 投入。

The purpose of this paper is the identification of high-fidelity digital twin data models from numerical code outputs by non-intrusive techniques (i.e., not requiring Galerkin projection of the governing equations onto the reduced modes ba…

Liver cancer is a complex disease responsible for a high number of deaths across the globe each year, making automated solutions for liver cancer classification urgent. The most common form of liver cancer is hepatocellular carcinoma (HCC)…

A Semgrep detector plus hand adjudication over 120 locally generated Python/TypeScript functions: syntax can surface 'silent failure' candidates, but the verdict lives outside the code.

华为在 2026 AIDC 产业发展大会暨 3D 数据中心现场会上发布全球首个 3D 数据中心，并发行《3D 数据中心》专著、开放《3D 数据中心概要设计图库》。该数据中心采用供冷层、IT 设备层、供电层、备电层四层垂直架构，具备分区隔离、故障不蔓延能力，机电交付时间减半。华为云已在贵州贵安、内蒙古和林格尔、安徽芜湖三大云核心枢纽规模化建设 3D 数据中心。

Salesforce 在 Dreamforce 大会上发布首款推理大模型 Koa，基于英伟达开源权重的 Nemotron 模型由双方联合微调，主打销售、市场营销和客服类业务任务。Koa 将作为 Agentforce 平台可选模型，训练未使用真实客户数据，Token 消耗更低，可经 AI 网关自动调度；此前复杂推理任务依赖 Claude、ChatGPT 等外部前沿模型。

Multi-target regression requires a model to simultaneously predict several related outputs. Conformal prediction provides distribution-free, finite-sample marginal coverage guarantees, but extending these to joint multi-dimensional regions…

The choice of Riemannian metric can strongly influence the convergence of gradient-based optimization over covariance matrices. Euclidean, Bures-Wasserstein and affine-invariant metrics are common choices, but their relative effectiveness …

Recent advancements in large language models have significantly enhanced the capabilities of agents in modeling long-term conversations. Despite these successes, existing approaches typically adopt a static heuristic paradigm, where inform…

Grounded language-model systems are often evaluated by final answer accuracy, yet a correct answer can be unsupported, drawn from the wrong source, or produced when evidence is insufficient or contradictory. We introduce EviScope, a paired…

Few-shot audio classifiers may rely on foreground-background co-occurrences and fail when those correlations shift. On SpurAudio, the resulting representation shift is concentrated and class dependent: for ResNet12, the top 10 percent of c…

gamescom 2026 圆满收官！🎮 Business Area 展示 AI 3D 创作。Entertainment Area 展示 AI 原生玩法。 感谢交流、试玩和坦诚反馈--也感谢 @JupiterSr632 和 @Elegoo_Official 帮助把 Meshy 带到展会现场！❤️

华为云推出 HuaweiCloudStack，整合云、数据与 AI，构建面向智能时代的 Agentic Hybrid Cloud 基础架构。该方案旨在帮助企业应对 AI 智能体快速规模化带来的数据、算力与业务运营变革，将智能体 AI 能力转化为实际业务价值。

一份 13 页的 Agent Memory Architecture 白皮书把可学习、能自愈的智能体拆成 5 层：工作记忆（上下文窗口）、情景记忆（带时间戳日志）、语义记忆（知识图谱）、程序记忆（固化成功路径为可重用 Skill）和遗忘引擎（修剪矛盾数据）。

Congenital heart disease (CHD) diagnosis and surgical planning often require patient-specific 3D anatomical models, but manual segmentation is labor-intensive, particularly in complex anatomies. Although deep-learning methods can automate …

Indirectly encoded neural networks can assign different activation functions to individual nodes, but the right functions are rarely known in advance. When the available set contains only standard monotonic functions, problems like parity …

GPT Images 2.5 的表情参考功能可将人的表情迁移到猫猫、动物或人物上。用户只需输入"把图二猫咪变成图一那样的表情"这类提示词即可完成表情替换。该功能被评价为表情参考效果很好，适合制作表情包。

AI assistance places a metacognitive demand on users, who must judge their own competence and the system's. Yet designers lack comparative evidence on which interventions to choose, where to place them, and how to tell whether they worked.…

Assistive autonomous systems must anticipate human goals before an observed behavior is complete. This article formulates anticipation as goal inference from a partially observed multimodal episode together with structured prediction of th…

A few days ago, I posted a tutorial in the Spring AI Dev community on how to use Anthropic Claude’s...

Message-passing Graph Neural Networks (GNNs) iteratively propagate and aggregate local neighborhood information followed by global readout to learn graph representations. However, their discriminative power is upper-bounded by the Weisfeil…

Current predictive turn-taking models (PTTMs) achieve strong performance on benchmarks with controlled acoustic conditions and clean audio signals. Their generalisation to conversations with overlapping speech and background interference r…

看来还是 AI 听指令， 我跟同事也反复强调不要"过度设计"， 然并卵。

I call this an automated blog, and technically that's true -- Windows Task Scheduler wakes a sleeping...

Mozilla 9 月 15 日发布的 State of Open Source AI 报告称，美国闭源前沿模型与中国最佳开源权重模型的性能差距已缩至 4.4 个月，Kimi K3 在 Artificial Analysis Intelligence Index 上仅落后 Anthropic Fable 5 三分，成本只有其 30%。

9 月 10 日，可灵 AI 参加了 TIFF： The Market 2026，通过行业座谈、交流酒会和现场展示，与全球电影社群建立联系。 从长篇动画到真实制作流程，我们探讨了 AI 视频如何成为专业电影制作的一部分。 从戛纳到多伦多乃至更多地方，可灵 AI 持续参与各大电影节，助力推动 AI 视频在制作中的作用。

Salesforce 在 Dreamforce 大会上发布首个推理模型 Koa，基于 Nvidia 开源权重的 Nemotron 模型，经后训练擅长销售、营销和客服任务，将作为 Agentforce 平台中 Claude 或 ChatGPT 等前沿模型的替代选项。

We study nonconvex methods for matrix completion, the problem of recovering a low-rank matrix from a subset of its entries. Convex methods achieve sample complexity linear in the matrix dimension and the rank, up to logarithmic factors, wh…