AI 情报文章归档

Nearly 90% of drugs entering clinical development ultimately fail, despite billions of dollars in investment. Pharmaceutical companies therefore rely on clinical development planning (CDP) and probability of technical and regulatory succes…

Large language models (LLMs) have achieved state-of-the-art performance across a wide range of tasks, motivating two important aspects of deployment: inference efficiency and output provenance, which can be tackled by speculative sampling …

LLMs are increasingly used in social simulations for socially interactive agents and robots, offering more flexibility than rule-based systems. However, even though they mimic human behaviour very well, there is a persistent alienness to t…

A North Star Metric is one number that represents value delivered. That is the whole...

今天我们在 Runway 内推出统一计价，所有购买的 credits 均可在我们的 web 应用和 Runway Dev 中使用。 现在，一次购买即可支持你在任意偏好的空间中使用 Runway 的所有工作方式。同时，我们还推出了与之匹配的新优惠。

蒙特利尔 AI 行业活动期间，反 AI 抗议者走上街头，在 René-Lévesque 大道游行。抗议者称 AI 生成的传单和网络梗不值得消耗行业淡水资源，防暴警察阻止抗议者穿越该大道。

Engrams Embedding Entendre：面向高效 DRAM/SSD 卸载的协同设计 新模型架构对 DRAM/NVMe TAM 的影响，DeepSeek V4.1 Flash，AgentX，InferenceX，NVMe 实验 https://newsletter.semianalysis.com/p/engrams-embedding-entendre-codesign

On-device inference is booming, but the momentum is almost all in language models. Diffusion pipelines are memory hungry, latency-sensitive, and require orchestrating an embedder, a transformer, a decoder, and often further postprocessing …

I've been using AI coding agents heavily, and I kept running into the same problem: Useful...

Long transcripts are costly inputs for downstream NLP systems and often contain irrelevant context. We study query-conditioned topic localization: predicting the sentence span in a transcript that best addresses a topic-title query. To imp…

Frontier language models now produce professional deliverables that expert graders judge to match human work on a substantial share of economically valuable tasks, yet most enterprise GenAI initiatives fail to show a measurable business ef…

SemiAnalysis 分析 Engram 架构（在标准 token 嵌入上扩展学习式多 token 查找）如何通过协同设计将嵌入表卸载到主存 DRAM 或 SSD，在同等质量下降低模型所需的 HBM 容量。

Google DeepMind 的 Rohin Shah 和 Anca Dragan 在 Deepmind Institute 发文指出，可见的思维链（CoT）是关键安全优势，例如 Gemini 3 Pro 的思维链曾揭示模型意识到自己处于测试环境。

现在你可以用 GPT6-Astra 构建任何可玩的 Three.js 游戏 收集你的星光，打开月之门！

Clustering high-dimensional data is a fundamental task in unsupervised machine learning with applications to a variety of domains. In the centralized data scenario, this task is commonly solved using deep clustering methods that utilize de…

Blind and low-vision users often face challenges when locating and physically acquiring objects in unfamiliar indoor environments. Existing vision-language-model-based assistants can provide semantic descriptions but may introduce latency,…

Speculative decoding accelerates LLM inference by drafting multiple tokens in parallel, with tree-based methods further improving efficiency through hierarchical structures. Dynamic-tree methods such as EAGLE-3 perform well under greedy de…

长安汽车首席专家、长安天枢智能机器人公司总经理谭欢在泰达国际论坛具身智能专场提出，汽车未来形态是"汽车机器人"--自学习、自组织、自进化的组合智能体。长安的机器人路径是先把座舱、中控、扶手等部件组件化，再走向覆盖搬运、物流、清洁、储能、补能的智能出行生态机器人，并拓展巡检、安防、消防等极限场景。整体逻辑为用机器人造车、卖车、上车，最终实现用机器人造机器人。

This paper presents a novel framework integrating Bayesian physiological modeling with optimal control strategies to achieve uncertainty-aware, personalized hemodynamic regulation during fluid resuscitation. An uncertainty-aware variationa…

中东战争与石油出口下滑正威胁支撑 AI 基建的资本流入，但迄今尚未明显放缓。沙特本月承诺 150 亿美元本土 AI 投资，卡塔尔是唯一收缩的中东主权基金，阿联酋和沙特仍在加码。Meta 上周推出的个人 AI 助手 Muse 已对 Instinct 形成冲击，后者据称正寻求 100 亿美元估值。

Adaptive optimization methods such as AdaGrad and Adam are widely used in modern neural-network training, but their adaptive scaling is primarily designed for vector-valued parameters and does not explicitly exploit matrix structure. Recen…

Multimodal survival models can combine complementary prognostic information from whole-slide images and genomic profiles, but effective fusion remains challenging amid external cohort shift and computational complexity. To address these ch…

Background: Just-in-time adaptive interventions (JITAIs) can use behavioral data to adapt support to changing contexts, but many rely on predefined rules and manual configuration. Objective: We developed a proof-of-concept sleep JITAI usin…

We study how far a simple statistical pipeline can go on univariate time series anomaly detection under a strict selection protocol. The method extracts a small pool of statistics over sliding windows, scores each window with a transductiv…

AI Evaluator Forum 发表公开信，超 100 名 AI 专家联署，向前沿 AI 开发商提出嵌入式评估的五项最低要求，包括评估者编辑独立、多元评估者、公开运作条款、免遭报复和获得等同高权限员工的访问级别。

据路透社 9 月 18 日报道，Dario Amodei、Sam Altman 和马斯克先后警告 AI 风险并主张控制发展节奏，欧洲企业和官员对此明显存疑。Mistral、Proton、Black Forest Labs 等指责此举意在巩固美国巨头市场地位，Hugging Face CEO 也反对放慢，但赞同在企业内部设置独立评估人员。

Dwarkesh Patel 发布与 OpenAI 核心推理科学家 Noam Brown 的约 80 分钟访谈，讨论多智能体、纳维-斯托克斯方程、数学进展与递归自改进，以及如何在启动 RSI 前确认模型是否对齐。

Anthropic 发布透明度指标，称 Claude 在 AL0 到 AL5 的 Epoch AI 量表上以 AL4 级别主导 26% 的模型开发工作，高于 2 月的不足 1%。

Defenses against jailbreak attacks on Large Language Models (LLMs) operate at different pipeline stages, such as input modification or output guard, but it remains unclear which defenses to deploy at each stage and how to combine them. Pri…

As deep learning models continue to advance, knowledge tracing models have achieved higher accuracy. However, these gains come at the cost of reduced interpretability, which is crucial for practitioners in educational settings to adopt new…

Most multilingual dysarthria-severity systems either train on a single aetiology-language pair or pool heterogeneous aetiologies into one label space. We test that pooling assumption with four matched HuBERT-base contrastive embedding mode…

a16z 的 Charts of the Week 汇总多组数据指出，AI 代码生成工具推动 iOS、Android 和 Chrome 上每月新增应用翻倍至翻两番，但下载量（iOS 为评分数）基本停滞，达到 10+ 评分或 100+ 下载等 escape velocity 门槛的应用占比大幅下降，作者称之为 App-Slop。

A pretrained robot foundation policy may execute most of a long-horizon task yet repeatedly fail at a few critical subtasks. Collecting additional full-task demonstrations for supervised fine-tuning (SFT) requires operators to repeat behav…

Kevin Lu 在 @vllm_project 发了一篇很棒的博客，讲推理引擎的质量控制。遗憾的是，@AnushElangovan 没有为 vLLM 提供足够稳定的 AMD QA 集群，导致 AMD 上的软件质量比 CUDA 差了一个数量级。 每个月都会发生多次 AMD CI vLLM 全集群宕机。

Hacktron AI 三人团队在 OpenAI 漏洞赏金计划中利用 Claude Opus 5 发现并串联两个严重漏洞，接管多个 OpenAI 员工 ChatGPT 账户，获 6500 美元奖金，OpenAI 表示已修复。

Quick disclosure up front, because it matters for how much weight to put on anything below: this was...

Ten parts ago, ClinicLive worked but looked like the template it was born from. This series made the...

TechCrunch Disrupt 2026 的 Expo Hall 参展桌位预订将于今晚（9 月 18 日）11：59 p.m. PT 截止，此后创业公司无法再加入。

Google 扩展 AI & Economy 研究项目，新增诺贝尔经济学奖得主 Philippe Aghion 等学术顾问与访问学者，并任命 Anu Madgavkar 和 Daniel Rock 为项目研究总监。该项目聚焦未来工作、生产力增长、全球技术扩散及 AI 对科学发现的影响，新团队将直接为 ATLAS 后续更新和实证研究提供支持。

Introduction You build a dispatch command centre with ToolJet MCP for a regional broadband...