AI 情报文章归档

Microsoft AI 发布 Sequential Diagnosis Benchmark（SDBench）和 AI 诊断系统 MAI-DxO，论文发表于 Science。

Microsoft Bing has introduced Topics Portfolio, a beta feature in Bing Webmaster Tools that helps...

Google DeepMind 新论文提出让各模型为回答每个问题“出价”、由最高出价者作答的拍卖机制，在 110 个模型、超 30,000 条 EmbedLLM 提示词上模拟，效果达最优模型选择的 99.2%，而按提示词类型路由到最佳模型仅 89.4%。

试试 Grok @Bot

Microsoft 在 Science 发表研究，称其 Diagnostic Orchestrator 在 SDBench 基准上诊断准确率最高达人类医生的四倍，计划开源该基准。

Robotic foundation models are typically evaluated through costly closed-loop rollouts, which motivates cheap offline proxies for model capability. The Platonic Representation Hypothesis (PRH) suggests that as representations improve, diffe…

Test-time scaling has emerged as a key inference strategy for reasoning models, where a model generates a large number of candidate solutions in the hope of finding at least one correct solution. Running all candidate solutions to completi…

Constructing accurate and computationally tractable approximation spaces is a fundamental task in scientific computing. Partially trainable neural networks bridge fixed random features and fully adaptive representations. This paper quantif…

We consider online monitoring of an AI system that produces a sequence of decisions. At each round, the system outputs a proposed decision together with a confidence score estimating the trustworthiness of the decision. The decision and co…

Auditing whether a text-to-image (T2I) model was fine-tuned on a particular image collection is difficult: a model can absorb collection-level visual characteristics without reproducing training images. Proactive dataset-use auditing plant…

Wrong automatic routing of an IT support ticket can cost more than human triage. We evaluate cost-aware abstention as an offline, cost-conditional readiness assessment: select a routing policy on calibration data under explicit costs, then…

Meta-reinforcement learning aims to train agents that adapt quickly to unseen tasks from limited interaction, but this requires representing task evidence over long interaction histories. Classical recurrent agents compress experience into…

Tool-using language model agents rely on environmental facts to decide whether an action is permitted. When a tool response omits a safety-relevant fact, an agent may carry out a prohibited action even though the user request is benign and…

Many recent open-weight language models use n-gram embedding tables to improve performance. These tables store learned representations of short token sequences, such as pairs or triples of consecutive tokens, which the model retrieves duri…

Anthropic 宣布将 Agent Skills 捐赠给 Agentic AI Foundation（@AgenticAIFdn），此前 MCP 也捐赠给了同一基金会。自 2025 年 12 月以来，已有 70 多个智能体平台接入该标准，包括 ChatGPT、Cursor、GitHub Copilot 和 VS Code。

Standard implementations of softmax, log-softmax, and cross-entropy have two numerical bugs. The first bug, Softmax Collapse, rounds the target logit's gradient to zero if that one logit is much larger than the others, resulting in gradien…

Chain-of-thought (CoT) prompting elicits step-by-step rationales that appear to explain model predictions, but there is no guarantee these steps causally drive the final answer. In standard autoregressive generation, dense attention allows…

Multilingual pipelines such as FineWeb2 apply one recipe across a thousand languages, leaving the failure modes of individual languages unanticipated. We use an LLM once, as a corpus debugger rather than a classifier: a tool-using agent ex…

Google 宣布推出 Agent Compute Engine，为每个 AI 智能体分配独立虚拟机，已支撑 Gemini，将在未来几个月开放预览。每个智能体运行在由 Google Titanium 芯片保护的 microVM 中，空闲时 VM 自动暂停，共享内存降低运行成本，Google 称单台服务器可容纳十倍的智能体数量。

Looped architectures scale quality while keeping parameter count fixed by repeatedly passing activations through the same block of layers. During inference, each recurrence generates independent KV-caches, causing memory to grow linearly. …

OpenAI 总裁 Greg Brockman 宣布与 Uber 合作，将 ChatGPT Work 推向 Uber 全体员工，为全公司带来智能体能力。Uber 方面称此前已完成 1000 人试点，起步于多年前客服团队使用 4K 上下文窗口的 ChatGPT，如今用前沿模型在全球规模自动化工程工作流；其今年推出的 AI pods 中，一个团队将覆盖 150 个城市的资本分配流程从 15 小时缩短到 30 分钟。

PyTorch Foundation 公布 2026 届大使名单，共 24 位社区领袖，覆盖五个项目区域，从尼日利亚、布基纳法索到新加坡、印度、罗马尼亚、西班牙、英国和美国。

Instrumental convergence is often invoked as a general reason to expect capable agents to acquire or express power-seeking subgoals: self-preservation, resource acquisition, deception, or resistance to oversight which poses an important th…

While monolingual language model (LM) surprisal is widely established as a predictor of reading times, it cannot capture the non-native (L2) reading dynamics that are shaped by the reader's first language (L1). In this work, we investigate…

如今AI实验室都在承受两年前做出的算力押注，而两年前没有哪家公司能预见AI算力需求会变得多大。 > “我不认为今天世界上会有哪家公司坐下来跟你说，你知道，他们两年前的决定做对了。” ——桑达尔·皮查伊 ---- 完整视频见“Sources Podcast, Alex Heath (@alexeheath) and Sundar Pichai”YouTube频道

On-policy distillation has been widely adopted for alleviating the distribution mismatch of off-policy distillation. Self-distillation further removes the need for an external teacher by using the student's own outputs, enhanced with extra…

Latent Space 在 Modal 完成 7.5 亿美元 Series D（估值 160 亿美元、年化收入超 10 亿美元）之际，专访 CEO Erik Bernhardsson，解释支撑增长的基础设施与经济逻辑。

We study learning a discrete distribution under person-level differential privacy in the batched setting. Given $n$ users, each contributing $m$ i.i.d. samples from an unknown distribution $μ$ over $[k]$, we seek an estimate of $μ$ with to…

Atlas 让你从单张输入图像开始构建大型 3D 环境。 随后你可以在该世界中定义一条相机路径，并沿路径生成视频。 下面展示了 3D 结构、相机路径与最终飞越视频之间的关联。

Purpose: To describe an evidence-aware retrieval evaluation protocol for a clinical retrieval-augmented generation (RAG) system over breast lesion management literature, addressing the dependence of retrieval scores on annotation coverage.…

Google Gemini now provides two first-party ways to turn a natural-language prompt into a working web...

Liquid AI 发布实时推理管线 RIPL，可在实时请求路径中服务 230M 至 2.6B 参数的 LFM2.5 模型，端到端生成响应低于 50 ms。按其自测数据，RIPL 的 p95 延迟比 vLLM 上的 Qwen3.5 低 5.6 至 28.4 倍，其中 LFM2.5-1.2B 为 155.8 ms 对 4,431 ms。

来花我们的 token OpenClaw focus weeks 邀请你加入我们的团队服务器，与我们一起构建 你可以使用我们的智能体、花我们的 token，并获得我们维护者的专业建议和帮助 加入前沿 https://openclaw.ai/blog/openclaw-focus-weeks

@JensenHuang 主题演讲对 https://pollen-robotics.com/microduck 预售的影响。 明天总预订单量可能会突破 20,000+，这或许会让 Microduck 成为 2026 年销量最高的机器人？

Counterfactual explanation methods have been introduced for unimodal settings, with dedicated frameworks for tabular, timeseries, text, and image data, yet no post-hoc method has been designed for classifiers that learn from multimodal inp…

Control evaluations measure whether safeguards can prevent an AI model from causing undesired outcomes even when it is intentionally trying to do so. These evaluations typically use purpose-built control settings and extrapolate their resu…

使用经 AA-PersonalityBench 评估的人格特质，构建你自己的自定义人格套件

早上6点的我，带着你的每日简报

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady ...

外面的情况相当糟糕（网络攻击方面）。 人们抱怨 AI 写出垃圾代码。但我们的软件和互联网基础设施，靠的是人类 50 年来写下的垃圾代码在运转。 如今连笨拙的 AI 模型都在变成强大的黑客工具。 竞赛已经开始：防守方需要用 AI 抢在攻击者之前找到并修补漏洞。