🤖 AI 速览

今天的主线是 AI 从通用助手继续下沉到专业服务与真实基础设施。OpenAI 推出法律场景方案，Anthropic、Notion强化项目级与团队级工作流；同时，数据中心、消费端入口与多篇可信系统论文提示，下一阶段竞争不只在模型能力，更在证据溯源、权限治理、长期记忆和可持续成本。
📋 文章元数据
发布时间
2026-09-18
类型
ai-daily
字数
6111
阅读时长
29 min

2026-09-18 AI日更 | OpenAI切入法律场景：AI落地开始拼专业数据、证据链与算力成本 链接到标题

今天的主线是 AI 从通用助手继续下沉到专业服务与真实基础设施。OpenAI 推出法律场景方案，Anthropic、Notion强化项目级与团队级工作流；同时，数据中心、消费端入口与多篇可信系统论文提示，下一阶段竞争不只在模型能力，更在证据溯源、权限治理、长期记忆和可持续成本。

📖 本期 Watch List 深度导读 链接到标题

今天最值得放进 Watch List 的，是三条线索。首先是“AI 基础设施与普通用户落地”：Meta 数据中心与就业叙事、Joanna Stern 对 iPhone 与大众 AI 的讨论，分别从供给侧和消费侧观察 AI 下一阶段扩散。其次，几篇 arXiv 都指向“可信 AI 系统”：可独立挑战的 AI 辅助声明、GraphEcho 的证据溯源、文档版本治理与传感器数字孪生，适合关注企业级 Agent 风控的团队细读。第三是“数据与记忆能力”：NeMo Data Designer、多模态合成数据、CapMem 自我中心视频记忆，以及医疗、交易 Agent 场景论文，展示模型正从生成能力走向长期记忆、可验证决策和垂直执行。

🌐 X 平台 AI 热点快讯 链接到标题

话题 1:Anthropic Launches Redesigned Projects for Claude Code 链接到标题

  • 分类:AI · News
  • 概况:热度时间:6 hours ago,相关帖子数:3900
  • 是什么事:Anthropic 推出了面向 Claude Code 的新版 Projects 功能，旨在更好地组织代码上下文、任务和协作流程。
  • 为什么重要:这表明 AI 编程助手正从单次对话式工具走向项目级工作空间，提升模型在长期代码理解、上下文管理和软件开发流程中的实用性。
  • 讨论概况:X 上讨论主要集中在新版 Projects 是否能显著改善 Claude Code 的开发体验、与 Cursor/GitHub Copilot 等工具相比的竞争力，以及项目级上下文管理在真实工程中的可靠性和成本。

话题 2:Anthropic Unifies Claude Chat and Cowork into Single Seamless Experience 链接到标题

  • 分类:AI · News
  • 概况:热度时间:1 day ago,相关帖子数:11000
  • 是什么事:Anthropic 将 Claude 的聊天功能与协作工作场景整合为一个更无缝的统一体验。
  • 为什么重要:这显示 AI 助手正从单纯对话工具转向可深度参与工作流的协作型生产力平台，可能影响企业 AI 应用和智能体产品的发展方向。
  • 讨论概况:X 上讨论集中在新体验是否能提升实际办公效率、Claude 相比 ChatGPT 和 Gemini 的差异化优势，以及更深度接入工作流可能带来的数据隐私、权限管理和企业采用门槛。

话题 3:Z.ai’s AI Agent Triples Throughput on Domestic Hardware in Two Weeks 链接到标题

  • 分类:AI · News
  • 概况:热度时间:15 hours ago,相关帖子数:2900
  • 是什么事:据报道，Z.ai 的 AI Agent 在两周内通过优化将国产硬件上的吞吐量提升至原来的三倍。
  • 为什么重要:这表明模型与系统优化可能显著释放国产 AI 硬件的性能，有助于降低对海外硬件的依赖并提升本土算力生态的竞争力。
  • 讨论概况:X 上的讨论主要集中在三倍提升的测试口径、硬件与软件优化的具体贡献、能否在不同任务和规模下复现，以及这一成果对国产芯片性能和商业化落地的实际意义。

话题 4:Ex-OpenAI Engineer Launches Jev for Fast AI Decisions 链接到标题

  • 分类:AI · News
  • 概况:热度时间:2 days ago,相关帖子数:72000
  • 是什么事:前 OpenAI 工程师推出名为 Jev 的新项目，主打让 AI 系统以更快速度完成决策。
  • 为什么重要:如果 Jev 能显著降低 AI 决策延迟，将有助于推动智能体、多步骤自动化和实时应用落地，提升 AI 在金融、客服、编程和企业流程中的实用性。
  • 讨论概况:X 上讨论集中在 Jev 是否真的具备远超现有方案的速度优势、其技术细节是否透明、能否在准确性与安全性之间取得平衡，以及前 OpenAI 人才创业是否会继续加剧 AI 智能体赛道竞争。

话题 5:Notion Launches Skills API for Team AI Prompts 链接到标题

  • 分类:AI · News
  • 概况:热度时间:6 hours ago,相关帖子数:377
  • 是什么事:Notion 推出 Skills API，允许团队将可复用的 AI 提示词和工作流能力集成到 Notion AI 中。
  • 为什么重要:这表明 AI 产品正从单次对话走向可复用、可管理的团队级工作流，有助于企业把提示工程、知识库和业务流程标准化。
  • 讨论概况:X 上讨论集中在企业 AI 工作流是否会因 Skills API 更易落地；有人看好 Notion 借此增强团队协作和自动化能力，也有人担心提示词治理、权限安全、结果可靠性以及对特定平台的依赖。

话题 6:Dreamforce Spotlights AI Integrations and Safety Debates with Top Executives 链接到标题

  • 分类:AI · News
  • 概况:热度时间:2 days ago,相关帖子数:23000
  • 是什么事:Salesforce Dreamforce 大会聚焦企业 AI 集成落地，并邀请多位高管讨论 AI 安全与治理议题。
  • 为什么重要:这反映出 AI 正从模型能力展示转向企业软件、业务流程和安全合规的规模化应用，相关平台的集成方式可能影响企业采用 AI 的速度与标准。
  • 讨论概况:X 上讨论主要集中在企业 AI 是否已具备实际 ROI、AI 代理和办公自动化的落地边界，以及大型科技公司在推动 AI 商业化的同时如何承担安全、隐私和治理责任。

今日 X 上的 AI 舆情小结 链接到标题

今天的舆论主线是，AI 正从单次对话和模型能力展示，转向项目级上下文、团队协作、可复用工作流与企业业务流程的深度集成，智能体和 AI 编程工具成为竞争重点。整体共识是，工作流整合、软件与硬件协同优化以及更低的决策延迟，有望提升 AI 的实际生产力并加速企业落地；但市场仍分歧于这些产品能否带来可持续的真实 ROI，以及不同厂商的体验和性能优势是否足以形成差异化。围绕 Z.ai“三倍吞吐”、Jev 的速度优势及各类新功能的实际效果，讨论普遍要求更透明的测试口径、可复现结果，并关注速度提升是否以准确性和可靠性为代价。潜在风险则集中在企业数据隐私、权限与提示词治理、智能体失控和责任归属，以及对单一平台的依赖，若安全合规和长期成本未被解决，AI 从演示走向规模化应用仍可能受阻。

💡 大佬观点(Influencer Insights) 链接到标题

今日大佬观点暂缺,推荐阅读 Watch List 深度内容。

📚 附录:今日 Watch List 更新源列表 链接到标题

时间窗口:最近 3 天;覆盖 22 个源;共 36 条更新

Lex Fridman Podcast (A_full) 链接到标题

  • #502 – Psychiatry, Insane Asylums, Mental Illness, ECT, Lobotomies, Freud & Jung
    • 发布时间:2026-09-17 08:40 北京时间
    • 摘要:【待翻译】- Andrew Scull is a historian of psychiatry.
      • See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc.
      • Wispr Flow: AI-powered voice dictation app.
      • Fin: AI agent for customer service.
      • LMNT: Zero-sugar electrolyte drink mix.
    • EN 要点:
      • Andrew Scull is a historian of psychiatry
      • Thank you for listening ❤ Check out our sponsors:
      • See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc
      • CONTACT LEX:

All-In Podcast (A_full) 链接到标题

  • Meta’s Dina Powell McCormick: The Case for Data Centers, Backlash, AI Job Boom & Meta’s Future

    • 发布时间:2026-09-18 00:54 北京时间
    • 摘要:【待翻译】- IREN is a vertically integrated AI Cloud platform, delivering data centers, compute and software for AI training and inference.
      • Oracle connects the data, applications, and infrastructure that turn AI into business outcomes—with the flexibility, choice, and control to optimize as AI evolves.
      • EY helps tech innovators scale from startup to exit to megacap.
      • We’re focused on giving every person the tools to reach their full potential and making sure the benefits of technology are distributed to all.
      • Keel Infrastructure owns the power, land, and connectivity that HPC and AI run on - backed by secured energy assets and established grid interconnections across North America.
    • EN 要点:
      • (0:00) Dina Powell McCormick joins The Besties
      • (1:52) Impact of Meta’s data center in Richland Parish, LA: $50K teacher checks, revitalizing a small town community
      • (11:52) How Meta is alleviating data center concerns: pollution, electricity, water, noise, aesthetics; Misinformation campaigns
      • (19:57) America’s Workforce Academy, the job growth in re-industrializing America

  • Brad Gerstner: No AI Bubble, Semis Eat the Nasdaq & AI’s Take Off Problem

    • 发布时间:2026-09-17 11:04 北京时间
    • 摘要:【待翻译】- IREN is a vertically integrated AI Cloud platform, delivering data centers, compute and software for AI training and inference.
      • Oracle connects the data, applications, and infrastructure that turn AI into business outcomes—with the flexibility, choice, and control to optimize as AI evolves.
      • EY helps tech innovators scale from startup to exit to megacap.
      • We’re focused on giving every person the tools to reach their full potential and making sure the benefits of technology are distributed to all.
      • Keel Infrastructure owns the power, land, and connectivity that HPC and AI run on - backed by secured energy assets and established grid interconnections across North America.
    • EN 要点:
      • (0:00) Welcome Brad Gerstner
      • (1:01) Trump Accounts, Every Child a Capitalist & The CAC Scan
      • (5:07) Can AI revenue pay for the CapEx
      • (8:53) The Build Out Issue: Gigawatts, TAM, Token Growth, and Margin Expansion

Stratechery by Ben Thompson (A_full) 链接到标题

  • An Interview with Joanna Stern About the iPhone Duo and AI for Normal People
    • 发布时间:2026-09-17 18:00 北京时间
    • 摘要:【待翻译】- $15 / month or $150 / year.
      • Substantial analysis of the news of the day delivered via three weekly emails or podcasts.
      • Stratechery Interviews.
      • Interviews with leading public CEOs, private company founders, and discussions with fellow analysts.
      • A twice-weekly podcast from John Gruber and myself: 15 minutes an episode, not a minute less, not a minute more.
    • EN 要点:
      • An Interview with Joanna Stern About the iPhone Duo and AI for Normal People

OpenAI Blog (A_full) 链接到标题

  • Introducing Astra for Law
    • 发布时间:2026-09-17 08:00 北京时间
    • 摘要:【待翻译】- OpenAI for Law brings frontier intelligence for law, custom firm workflows, connected legal data sources, and legal-grade controls for confidential client work.
      • This piece from OpenAI Blog explains how Introducing Astra for Law shapes the broader AI and infrastructure landscape.
      • It also surfaces practical implications for founders, operators, and investors following Introducing Astra for Law.
    • EN 要点:
      • OpenAI for Law brings frontier intelligence for law, custom firm workflows, connected legal data sources, and legal-grade controls for confidential client work.

Lex Fridman (B_intro+search) 链接到标题

  • Psychiatry, Insane Asylums, Mental Illness, ECT, Lobotomies, Freud & Jung | Lex Fridman Podcast #502
    • 发布时间:2026-09-17 08:34 北京时间
    • 摘要:【待翻译】- Andrew Scull is a historian of psychiatry.
      • See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc.
      • Feedback - give feedback to Lex:.
      • AMA - submit questions, videos or call-in:.
    • EN 要点:
      • Andrew Scull is a historian of psychiatry
      • Thank you for listening ❤ Check out our sponsors:
      • See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc
      • Transcript:

ArXiv cs.AI (B_intro+search) 链接到标题

  • Making AI-Assisted Claims Independently Challengeable: Publication Authority and a Protocol for Falsifiable Publication Records

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17631v1 Announce Type: new.
      • Abstract: AI-assisted claims can appear authoritative when evidence, analysis, human authorization, presentation, and correction history refer to different states.
      • Provenance, attestation, and transparency expose history but alone do not specify the publication transition examined here.
      • We develop Publication Authority as an exact-state, non-transferable, single-use publication capability and instantiate it in PAC-2026 (Publication-Accountability Calculus), a machine-readable AIJIM Protocol candidate.
    • EN 要点:
      • arXiv:2609.17631v1 Announce Type: new
      • Abstract: AI-assisted claims can appear authoritative when evidence, analysis, human authorization, presentation, and correction history refer to different stat…
      • Provenance, attestation, and transparency expose history but alone do not specify the publication transition examined here
      • We develop Publication Authority as an exact-state, non-transferable, single-use publication capability and instantiate it in PAC-2026 (Publication-Accountabili…

  • EvolveTrade: Experience-Driven Policy Refinement for Self-Evolving LLM Trading Agents

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17632v1 Announce Type: new.
      • Abstract: Large language model (LLM) trading agents can combine market data, news, and executable analysis, but their behavior is often controlled by static hand-written tool-use policies that are fixed before deployment.
      • This limits their ability to adapt how they gather evidence, invoke tools, verify signals, and manage risk under changing market regimes.
      • We introduce EvolveTrade, a self-evolving framework that treats the system prompt of a tool-using trading agent as a text-parameterized policy.
    • EN 要点:
      • arXiv:2609.17632v1 Announce Type: new
      • Abstract: Large language model (LLM) trading agents can combine market data, news, and executable analysis, but their behavior is often controlled by static han…
      • This limits their ability to adapt how they gather evidence, invoke tools, verify signals, and manage risk under changing market regimes
      • We introduce EvolveTrade, a self-evolving framework that treats the system prompt of a tool-using trading agent as a text-parameterized policy

  • One Color Preprocessing Improves DSATUR

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17633v1 Announce Type: new.
      • Abstract: The Graph Coloring Problem (GCP) is NP-hard and DSATUR stands as one of the fastest heuristics for it despite producing colorings that typically use more colors than state-of-the-art coloring algorithms.
      • We propose SSLD (Semidefinite Spectral Learning with DSATUR), which improves DSATUR by preprocessing a first good color class before letting DSATUR complete coloring the rest of the given graph.
      • We obtain this color class from a Semidefinite Programming (SDP), similar to an SDP used to compute the Lov'asz theta number.
    • EN 要点:
      • arXiv:2609.17633v1 Announce Type: new
      • Abstract: The Graph Coloring Problem (GCP) is NP-hard and DSATUR stands as one of the fastest heuristics for it despite producing colorings that typically use m…
      • We propose SSLD (Semidefinite Spectral Learning with DSATUR), which improves DSATUR by preprocessing a first good color class before letting DSATUR complete col…
      • We obtain this color class from a Semidefinite Programming (SDP), similar to an SDP used to compute the Lov'asz theta number

  • Physics-Constrained Digital Twins for Sensor Integrity in Urban Pedestrian Flow: Detecting Stealthy False Data Injection with Conformal Guarantees

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17635v1 Announce Type: new.
      • Abstract: City pedestrian counting systems now feed economic indicators, planning decisions and safety operations, yet the twins built on top of them treat the incoming stream as ground truth.
      • We study what happens when it is not.
      • We formalise stealthy false data injection for city-scale pedestrian sensing, where the map from latent flow to observation is far more rank deficient than in the power and water networks for which stealth has been characterised.
    • EN 要点:
      • arXiv:2609.17635v1 Announce Type: new
      • Abstract: City pedestrian counting systems now feed economic indicators, planning decisions and safety operations, yet the twins built on top of them treat the…
      • We study what happens when it is not
      • We formalise stealthy false data injection for city-scale pedestrian sensing, where the map from latent flow to observation is far more rank deficient than in t…

  • What You Can’t See Is Still What You Learn: A Preregistered Sixty-Society Confirmation That Evidence Masking Drives Compositional Generalization

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17637v1 Announce Type: new.
      • Abstract: Restricting what a module can read may improve what a system learns to compute.
      • We test this in a preregistered confirmation with sixty four-cell systems sharing a frozen language-model backbone and communicating through learned continuous packets.
      • Five conditions vary evidence masking, ownership markers, and replacement of foreign evidence with neutral filler, across six initialization clusters, each with two data orders, on one fresh task world.
    • EN 要点:
      • arXiv:2609.17637v1 Announce Type: new
      • Abstract: Restricting what a module can read may improve what a system learns to compute
      • We test this in a preregistered confirmation with sixty four-cell systems sharing a frozen language-model backbone and communicating through learned continuous…
      • Five conditions vary evidence masking, ownership markers, and replacement of foreign evidence with neutral filler, across six initialization clusters, each with…

  • CapMem: A Benchmark for Caption-Based Episodic Memory in Egocentric Video

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17688v1 Announce Type: new.
      • Abstract: Wearable assistants require episodic memory over egocentric video, yet current vision-language models face bounded frame budgets, growing visual-token costs, and long-context retrieval failures.
      • Under these practical constraints, we study whether textual captions can serve as reusable episodic memory.
      • We define the Episodic Memory Video Caption QA task and introduce CapMem, a human-annotated benchmark with 75 videos totaling 33.7 hours, and 1,000 multiple-choice questions across 16 scenarios.
    • EN 要点:
      • arXiv:2609.17688v1 Announce Type: new
      • Abstract: Wearable assistants require episodic memory over egocentric video, yet current vision-language models face bounded frame budgets, growing visual-token…
      • Under these practical constraints, we study whether textual captions can serve as reusable episodic memory
      • We define the Episodic Memory Video Caption QA task and introduce CapMem, a human-annotated benchmark with 75 videos totaling 33.7 hours, and 1,000 multiple-cho…

  • GraphEcho: Structural Redundancy and Evidence Provenance in LLM Graph Agents

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17695v1 Announce Type: new.
      • Abstract: A large language model (LLM) agent can follow more graph paths without acquiring more independent evidence.
      • GraphEcho tests whether agents mistake these repeated encounters for additional corroboration.
      • The benchmark varies path counts and evidential origins while holding evidence content fixed, and evaluates both judgments and active exploration.
    • EN 要点:
      • arXiv:2609.17695v1 Announce Type: new
      • Abstract: A large language model (LLM) agent can follow more graph paths without acquiring more independent evidence
      • GraphEcho tests whether agents mistake these repeated encounters for additional corroboration
      • The benchmark varies path counts and evidential origins while holding evidence content fixed, and evaluates both judgments and active exploration

  • GVD: Governed Versioning and Deduplication for Document Repositories

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17696v1 Announce Type: new.
      • Abstract: Document repositories evolve continuously.
      • Guidelines and policies are revised, superseded, and re-uploaded, so the same content recurs in different wording and newer versions refine or contradict earlier ones.
      • These inconsistencies belong to the growing collection rather than to any single document, yet existing work treats versioning, duplicate detection, and contradiction detection as isolated pairwise tasks and stops once a pair is labeled.
    • EN 要点:
      • arXiv:2609.17696v1 Announce Type: new
      • Abstract: Document repositories evolve continuously
      • Guidelines and policies are revised, superseded, and re-uploaded, so the same content recurs in different wording and newer versions refine or contradict earlie…
      • These inconsistencies belong to the growing collection rather than to any single document, yet existing work treats versioning, duplicate detection, and contrad…

  • NeMo Data Designer: An Extensible Framework for Multimodal Synthetic Data Generation

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17699v1 Announce Type: new.
      • Abstract: We present NeMo Data Designer (NDD), an open-source, general-purpose framework for multi-modal synthetic data generation (SDG).
      • Designed to be intuitive to use, NDD provides a declarative configuration format in which human and/or agent users define each dataset column, with column types spanning text, code, structured outputs, images, embeddings, and statistical samplers that are explicitly configured to steer dataset diversity.
      • Additional column types and functionality can be introduced using the framework’s flexible plugin system.
    • EN 要点:
      • arXiv:2609.17699v1 Announce Type: new
      • Abstract: We present NeMo Data Designer (NDD), an open-source, general-purpose framework for multi-modal synthetic data generation (SDG)
      • Designed to be intuitive to use, NDD provides a declarative configuration format in which human and/or agent users define each dataset column, with column types…
      • Additional column types and functionality can be introduced using the framework’s flexible plugin system

  • A Systematic Evaluation of the COTQ Provincial Land Cover Product: Structural Consistency, Spectral Separability, and Relative Positioning Against ESA, ESRI, and Google Products

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17731v1 Announce Type: new.
      • Abstract: High-resolution land use and land cover (LULC) products derived from Sentinel-2 imagery are widely used for environmental monitoring and land management, yet their performance can vary across regions with complex ecological gradients and heterogeneous surface conditions.
      • In Quebec, these limitations motivated the development of a provincial 10-m land-cover product, the COTQ, designed to support annual monitoring of land occupation and soil artificialisation.
      • This study presents a systematic evaluation of the COTQ product relative to three global 10-m LULC datasets: ESA WorldCover, ESRI LandCover, and Google DynamicWorld.
    • EN 要点:
      • arXiv:2609.17731v1 Announce Type: new
      • Abstract: High-resolution land use and land cover (LULC) products derived from Sentinel-2 imagery are widely used for environmental monitoring and land manageme…
      • In Quebec, these limitations motivated the development of a provincial 10-m land-cover product, the COTQ, designed to support annual monitoring of land occupati…
      • This study presents a systematic evaluation of the COTQ product relative to three global 10-m LULC datasets: ESA WorldCover, ESRI LandCover, and Google DynamicW…

ArXiv cs.CL (B_intro+search) 链接到标题

  • Enhancing Extubation Failure Prediction with LLM-Derived Features from Respiratory Therapy Clinical Notes

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17532v1 Announce Type: new.
      • Abstract: Invasive mechanical ventilation is a lifesaving therapy, but timely, safe discontinuation is essential to preventing extubation failure (EF) and related risks to health.
      • We present a novel approach to EF prediction that leverages features classified in free-text respiratory therapy notes using a large language model and logistic regression pipeline.
      • Applied to a patient cohort from University of Washington Medicine, our method identifies clinically meaningful EF-related features that improve EF prediction performance when included alongside structured patient data.
    • EN 要点:
      • arXiv:2609.17532v1 Announce Type: new
      • Abstract: Invasive mechanical ventilation is a lifesaving therapy, but timely, safe discontinuation is essential to preventing extubation failure (EF) and relat…
      • We present a novel approach to EF prediction that leverages features classified in free-text respiratory therapy notes using a large language model and logistic…
      • Applied to a patient cohort from University of Washington Medicine, our method identifies clinically meaningful EF-related features that improve EF prediction p…

  • Faking Good and Faking Bad in LLMs: Response Distortion Across Dark Triad Personality Traits

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17534v1 Announce Type: new.
      • Abstract: Social desirability and impression management are pervasive sources of response distortion in human personality assessment, yet their effects on Large Language Models (LLMs) remain underexplored.
      • This study investigates whether contemporary LLMs systematically modulate the expression of Dark Triad traits (Machiavellianism, narcissism, and psychopathy) under fake-good and fake-bad conditions.
      • Seven state-of-the-art models were evaluated across two ecologically relevant contexts: employment selection and forensic evaluation, in which socially desirable or undesirable incentives were conveyed through contextual framing.
    • EN 要点:
      • arXiv:2609.17534v1 Announce Type: new
      • Abstract: Social desirability and impression management are pervasive sources of response distortion in human personality assessment, yet their effects on Large…
      • This study investigates whether contemporary LLMs systematically modulate the expression of Dark Triad traits (Machiavellianism, narcissism, and psychopathy) un…
      • Seven state-of-the-art models were evaluated across two ecologically relevant contexts: employment selection and forensic evaluation, in which socially desirabl…

  • DANTINOX: A Unified Framework for Multi-Paradigm Language Modeling

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17535v1 Announce Type: new.
      • Abstract: Language generation research increasingly spans three paradigms: autoregressive decoding, discrete masked diffusion, and continuous flow-matching.
      • Comparing them is difficult because each lives in a separate codebase, so measured differences often reflect implementation details rather than the paradigms themselves.
      • We present DantinoX, an open-source JAX/Flax library in which a single modular Transformer backbone serves all three paradigms.
    • EN 要点:
      • arXiv:2609.17535v1 Announce Type: new
      • Abstract: Language generation research increasingly spans three paradigms: autoregressive decoding, discrete masked diffusion, and continuous flow-matching
      • Comparing them is difficult because each lives in a separate codebase, so measured differences often reflect implementation details rather than the paradigms th…
      • We present DantinoX, an open-source JAX/Flax library in which a single modular Transformer backbone serves all three paradigms

  • Think Before You Comfort: Reflective Cognitive Alignment for Protocol-Grounded Elderly Stimulation Agents

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17536v1 Announce Type: new.
      • Abstract: Cognitive Stimulation Therapy (CST) offers non-pharmacological support for elders with cognitive impairment, yet scalability remains constrained by reliance on trained facilitators and severe data scarcity, particularly for privacy-sensitive, low-resource languages such as Cantonese.
      • While Large Language Models (LLMs) show promise for automated companionship, they often struggle to balance empathetic engagement with adherence to cognitive stimulation guidelines.
      • We propose a framework addressing these challenges along two complementary axes.
    • EN 要点:
      • arXiv:2609.17536v1 Announce Type: new
      • Abstract: Cognitive Stimulation Therapy (CST) offers non-pharmacological support for elders with cognitive impairment, yet scalability remains constrained by re…
      • While Large Language Models (LLMs) show promise for automated companionship, they often struggle to balance empathetic engagement with adherence to cognitive st…
      • We propose a framework addressing these challenges along two complementary axes

  • Relation Before Entity: Deferred Commitment in Language Model Factual Recall

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17537v1 Announce Type: new.
      • Abstract: We ask whether relation-type information (e.g., capital-of) and entity-specific information (e.g., France to Paris) become causally active at the final-token position at the same depth during recall.
      • Using four complementary causal diagnostics across four decoder-only models and eight prompt families, we find a robust temporal asymmetry: relation information becomes generation-controlling before entity information does.
      • Relation onset precedes entity onset by 10-16 tested layers (31-44% of network depth) at threshold 0.4, with the ordering holding across all 16 model-threshold combinations for thresholds 0.2-0.5.
    • EN 要点:
      • arXiv:2609.17537v1 Announce Type: new
      • Abstract: We ask whether relation-type information (e.g., capital-of) and entity-specific information (e.g., France to Paris) become causally active at the fina…
      • Using four complementary causal diagnostics across four decoder-only models and eight prompt families, we find a robust temporal asymmetry: relation information…
      • Relation onset precedes entity onset by 10-16 tested layers (31-44% of network depth) at threshold 0.4, with the ordering holding across all 16 model-threshold…

  • From Pixels to Pairs: A Comprehensive Benchmark of LLM-Based Key-Value Extraction in Noisy Document Settings

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17538v1 Announce Type: new.
      • Abstract: Large language models (LLMs) are increasingly used for structured information extraction from documents, yet their behavior under realistic OCR noise remains poorly understood.
      • We present a systematic benchmark of open-source instruction-tuned LLMs for key-value pair (KVP) extraction under both clean-text and noisy OCR conditions.
      • We evaluate representative decoder-only models (Gemma, Mistral, Qwen2.5, LLaMA 3, and DeepSeek) on the FUNSD, CORD, and SROIE benchmarks using both Gold-text annotations and OCR outputs from PaddleOCR, EasyOCR, and Tesseract.
    • EN 要点:
      • arXiv:2609.17538v1 Announce Type: new
      • Abstract: Large language models (LLMs) are increasingly used for structured information extraction from documents, yet their behavior under realistic OCR noise…
      • We present a systematic benchmark of open-source instruction-tuned LLMs for key-value pair (KVP) extraction under both clean-text and noisy OCR conditions
      • We evaluate representative decoder-only models (Gemma, Mistral, Qwen2.5, LLaMA 3, and DeepSeek) on the FUNSD, CORD, and SROIE benchmarks using both Gold-text an…

  • MudawanSn: A Gold-Standard Wolof-Arabic Parallel Corpus for Machine Translation

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17539v1 Announce Type: new.
      • Abstract: We present MudawanSn, a gold-standard resource of 1,271 sentence-aligned pairs manually translated from Wolof into Modern Standard Arabic (MSA).
      • The source texts are drawn from the MasakhaNER corpus and cover politics, society, religion, and sports in Senegalese news discourse.
      • Although multilingual resources such as FLORES-200 and NTREX include both Wolof and Arabic, no publicly available parallel corpus is specifically designed for the Wolof-Modern Standard Arabic language pair.
    • EN 要点:
      • arXiv:2609.17539v1 Announce Type: new
      • Abstract: We present MudawanSn, a gold-standard resource of 1,271 sentence-aligned pairs manually translated from Wolof into Modern Standard Arabic (MSA)
      • The source texts are drawn from the MasakhaNER corpus and cover politics, society, religion, and sports in Senegalese news discourse
      • Although multilingual resources such as FLORES-200 and NTREX include both Wolof and Arabic, no publicly available parallel corpus is specifically designed for t…

  • Register Bias in Complexity-Based Large Language Model Routing

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17542v1 Announce Type: new.
      • Abstract: Large language model services increasingly route each query to one of several models of differing capability, using a cheap estimate of query complexity to send easy queries to small models and hard queries to large ones.
      • I show that this routing step is not register neutral: text written in a non-standard English register, African American English or the English of second-language writers, is systematically assigned a lower-capacity tier than a meaning-equivalent standard-English version of the same query.
      • The effect is driven by a specific, common routing signal, input length, because non-standard registers omit function words and thus look shorter and therefore simpler; other complexity signals do not carry it.
    • EN 要点:
      • arXiv:2609.17542v1 Announce Type: new
      • Abstract: Large language model services increasingly route each query to one of several models of differing capability, using a cheap estimate of query complexi…
      • I show that this routing step is not register neutral: text written in a non-standard English register, African American English or the English of second-langua…
      • The effect is driven by a specific, common routing signal, input length, because non-standard registers omit function words and thus look shorter and therefore…

  • Large Language Models Versus Physicians in Traditional Chinese Medicine: A Real-World Clinical Case Evaluation

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17544v1 Announce Type: new.
      • Abstract: Large language models (LLMs) are increasingly being explored for clinical applications, yet their assessment for real-world traditional Chinese medicine (TCM) practice remains limited We constructed a clinical case library comprising 349 de-identified outpatient cases from 62 hospitals and evaluated 16 LLMs and a comparator cohort of 60 practicing TCM physicians using 60 representative cases selected from this library.
      • Model outputs and physician reports were anonymized and scored by five senior TCM experts across nine diagnostic and therapeutic dimensions.
      • Cutting-edge general-purpose LLMs achieved higher expert scores than the physician comparators, particularly for medical advice, treatment principles and selected diagnostic tasks.
    • EN 要点:
      • arXiv:2609.17544v1 Announce Type: new
      • Abstract: Large language models (LLMs) are increasingly being explored for clinical applications, yet their assessment for real-world traditional Chinese medici…
      • Model outputs and physician reports were anonymized and scored by five senior TCM experts across nine diagnostic and therapeutic dimensions
      • Cutting-edge general-purpose LLMs achieved higher expert scores than the physician comparators, particularly for medical advice, treatment principles and select…

  • Legal LLM Hallucination Should Be Evaluated as Failure of Legal Warrant

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17546v1 Announce Type: new.
      • Abstract: In this position paper, we argue that legal LLMs’ hallucinations should be evaluated as a failure of legal warrant rather than as factual inaccuracy or citation failure.
      • We define claim-authority warrant as the context-sensitive relation between a consequential legal claim and authority that exists, applies to the relevant jurisdiction, is current for the date of analysis, has the legal status represented by the system, and supports the proposition asserted.
      • Warranted legal generation is the broader system behavior that answers, narrows, asks, warns, corrects a false premise, or abstains according to that relation.
    • EN 要点:
      • arXiv:2609.17546v1 Announce Type: new
      • Abstract: In this position paper, we argue that legal LLMs’ hallucinations should be evaluated as a failure of legal warrant rather than as factual inaccuracy o…
      • We define claim-authority warrant as the context-sensitive relation between a consequential legal claim and authority that exists, applies to the relevant juris…
      • Warranted legal generation is the broader system behavior that answers, narrows, asks, warns, corrects a false premise, or abstains according to that relation

ArXiv cs.LG (B_intro+search) 链接到标题

  • Pay Only for Disagreement: Certified No-Regression Verdicts for Model Updates with Matching Label-Complexity Bounds

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17560v1 Announce Type: new.
      • Abstract: Every production model is updated, by retraining, fine-tuning, quantization, or a silent vendor swap, and each update risks being worse than what it replaced.
      • We formalize update promotion as certified paired risk-difference auditing.
      • Our starting point is a support identity: the risk difference between two models lives on the inputs where they disagree, observable without labels.
    • EN 要点:
      • arXiv:2609.17560v1 Announce Type: new
      • Abstract: Every production model is updated, by retraining, fine-tuning, quantization, or a silent vendor swap, and each update risks being worse than what it r…
      • We formalize update promotion as certified paired risk-difference auditing
      • Our starting point is a support identity: the risk difference between two models lives on the inputs where they disagree, observable without labels

  • Beyond Static RAG: An Adaptive, Tri-Metric Routing Framework for Efficient Long-Context Inference on Commodity GPUs

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17564v1 Announce Type: new.
      • Abstract: Deploying retrieval-augmented generation (RAG) on commodity GPUs such as the NVIDIA T4 (16 GB VRAM) exposes a practical failure mode we call the Compression Paradox: neural prompt compression can add key-value (KV) cache contention and preprocessing latency that outweigh generation-time savings, while skipping compression can cause out-of-memory (OOM) failures on long contexts.
      • We identify two distinct failure mechanisms when a vLLM-served LLM and a PyTorch-based compressor are co-deployed under tight memory budgets, and introduce the Tri-Metric Router, a deterministic, training-free policy that selects among Raw, Neural (LLMLingua-2), and Lexical (BM25) pipelines.
      • The router uses three CPU-side signals: spatial complexity ($L$), syntactic density ($\rho_{key}$), and type-token ratio (TTR).
    • EN 要点:
      • arXiv:2609.17564v1 Announce Type: new
      • Abstract: Deploying retrieval-augmented generation (RAG) on commodity GPUs such as the NVIDIA T4 (16 GB VRAM) exposes a practical failure mode we call the Compr…
      • We identify two distinct failure mechanisms when a vLLM-served LLM and a PyTorch-based compressor are co-deployed under tight memory budgets, and introduce the…
      • The router uses three CPU-side signals: spatial complexity ($L$), syntactic density ($\rho_{key}$), and type-token ratio (TTR)

  • Where Grokking Happens: Distributed Utility and Fourier Recoding Without a Module Switch

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17571v1 Announce Type: new.
      • Abstract: Where in a Transformer is the change from memorization to generalization functionally expressed?
      • We introduce Transition Games–behavior-aligned exact activation games with paired non-generalizing controls–and find distributed utility gain with a prospective block-0 attention bias; selected degree-two modes account for 67–92% of its addition contrast across replacement games, and a disjoint exact path study confirms that block-1 MLP mediates more of their effect than all other tested downstream paths in 12/12 pairs.
      • The sharper “MLP memorizes, attention generalizes” prediction instead reverses (-.331 bits/example at the memory anchor; 0/12 in the predicted direction), while routing onset, global rank collapse, and a prime-invariant architecture ridge also fail, identifying grokking here as spectral recoding of an existing distributed circuit rather than a module switch.
    • EN 要点:
      • arXiv:2609.17571v1 Announce Type: new
      • Abstract: Where in a Transformer is the change from memorization to generalization functionally expressed
      • We introduce Transition Games–behavior-aligned exact activation games with paired non-generalizing controls–and find distributed utility gain with a prospecti…
      • The sharper “MLP memorizes, attention generalizes” prediction instead reverses (-.331 bits/example at the memory anchor; 0/12 in the predicted direction), while…

  • Disentangling Algorithmic Bias from Archival Artifacts: A Controlled Audit of Vision-Language Model Valuation in Metropolitan Museum Archives

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17572v1 Announce Type: new.
      • Abstract: Auditing vision-language models (VLMs) for societal bias requires distinguishing direct algorithmic valuation disparities from confounders embedded within archival metadata.
      • In this study, we audit Contrastive Language-Image Pretraining (CLIP) models using historical artwork metadata from the Metropolitan Museum of Art Open Access collection (N = 1,500 total objects; N = 743 attributed works: Male n = 534, Female n = 209; n = 618 anonymous).
      • We establish a quantitative audit framework evaluating zero-shot CLIP logit differential scores across three semantic prompt pairs (masterpiece, quality, and influence).
    • EN 要点:
      • arXiv:2609.17572v1 Announce Type: new
      • Abstract: Auditing vision-language models (VLMs) for societal bias requires distinguishing direct algorithmic valuation disparities from confounders embedded wi…
      • In this study, we audit Contrastive Language-Image Pretraining (CLIP) models using historical artwork metadata from the Metropolitan Museum of Art Open Access c…
      • We establish a quantitative audit framework evaluating zero-shot CLIP logit differential scores across three semantic prompt pairs (masterpiece, quality, and in…

  • Temperon: Full-Time SAM Quality at a Third Less Wall-Clock

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17575v1 Announce Type: new.
      • Abstract: Sharpness-aware minimization (SAM) doubles the cost of every training step, yet its benefit concentrates where training ends.
      • We study where an expensive training mode should be spent and propose Temperon: a plain-SGD explorer for the first 43% of the epoch budget, then one scheduled hand-off that gives the entire final cosine anneal to a SAM-wrapped Muon refiner.
      • On CIFAR-10/100, SVHN and Tiny ImageNet (five seeds, times reported as epochs-to-target times an idle-GPU-calibrated epoch cost), Temperon matches the best full-time-SAM recipe on accuracy everywhere while reaching the hardest common target 35%, 34% and 32% sooner on three of the four, and sits a tier above the published SAM+SGD recipe at level cost.
    • EN 要点:
      • arXiv:2609.17575v1 Announce Type: new
      • Abstract: Sharpness-aware minimization (SAM) doubles the cost of every training step, yet its benefit concentrates where training ends
      • We study where an expensive training mode should be spent and propose Temperon: a plain-SGD explorer for the first 43% of the epoch budget, then one scheduled h…
      • On CIFAR-10/100, SVHN and Tiny ImageNet (five seeds, times reported as epochs-to-target times an idle-GPU-calibrated epoch cost), Temperon matches the best full…

  • When the Gradient Sees Rank: Provable Necessity, Causal Recruitment, and Composition in Trained Matrix Memories

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17594v1 Announce Type: new.
      • Abstract: Can gradient-based training learn the rank needed to store and compose associations in a matrix memory?
      • In our earlier study, we used a matrix-augmented reasoner on a task that admits a rank-1 solution, leaving this question open.
      • We train matrix memories on $K$ fresh key-value bindings whose exact linear recovery requires $\mathrm{rank}(Z) \geq K$.
    • EN 要点:
      • arXiv:2609.17594v1 Announce Type: new
      • Abstract: Can gradient-based training learn the rank needed to store and compose associations in a matrix memory
      • In our earlier study, we used a matrix-augmented reasoner on a task that admits a rank-1 solution, leaving this question open
      • We train matrix memories on $K$ fresh key-value bindings whose exact linear recovery requires $\mathrm{rank}(Z) \geq K$

  • Prior-Free Competitive Ratios for Improving Bandits: Scale, Curvature and Horizon Are Free, but Not Jointly Under Noise

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17595v1 Announce Type: new.
      • Abstract: In the improving multi-armed bandits problem, each of $k$ arms has an unknown nondecreasing, discretely concave reward curve $f_i$, and pulling arm $i$ for the $t$-th time yields $f_i(t)$.
      • For sufficiently long horizons, Blum and Ravichandran (ALT 2025) proved that randomized algorithms achieve an $O(\sqrt k)$ approximation to the best single arm when the scale $m=f^*(T)$ of the optimal arm is known ($T\ge2k$), and $O(\sqrt k\log k)$ when it is not ($T>4k$), against an $\Omega(\sqrt k)$ lower bound.
      • The logarithmic factor is unnecessary: a one-page \emph{probe-and-commit} algorithm achieves competitive ratio $4\sqrt3,\sqrt k$ for $T\ge2\lfloor\sqrt k\rfloor$, without any knowledge of the scale, and we determine the optimal ratio for every horizon, $\Theta(\sqrt k+k/T)$, also for unknown horizons.
    • EN 要点:
      • arXiv:2609.17595v1 Announce Type: new
      • Abstract: In the improving multi-armed bandits problem, each of $k$ arms has an unknown nondecreasing, discretely concave reward curve $f_i$, and pulling arm $i…
      • For sufficiently long horizons, Blum and Ravichandran (ALT 2025) proved that randomized algorithms achieve an $O(\sqrt k)$ approximation to the best single arm…
      • The logarithmic factor is unnecessary: a one-page \emph{probe-and-commit} algorithm achieves competitive ratio $4\sqrt3,\sqrt k$ for $T\ge2\lfloor\sqrt k\rfloo…

  • Lecture notes on Physics Informed Neural Networks, Neural Operators, and their applications

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17638v1 Announce Type: new.
      • Abstract: This is the set of lecture notes for the PhD course \href{ Informed Neural Network}, held at the University of Bozen/Bolzano} in the academic year 2025/2026.
      • The goal of the course was to introduce the concept of Physics Informed Deep Neural Networks (PINN) and Neural Operators (NOs), discuss their implementation from scratch in PyTorch and using advanced ad-hoc developed open-source libraries such as NVIDia PhysicsNeMo to address real-world problems in various fields (engineering, physics, petroleum reservoir).
      • We discuss recent topics such as Mixture-of-Models, Fourier Neural Operators, Physics-Informed Kolmogorov-Arnold Networks (PIKANs) and Fourier Neural Operators.
    • EN 要点:
      • arXiv:2609.17638v1 Announce Type: new
      • Abstract: This is the set of lecture notes for the PhD course \href{ Informed Neural Network}, held at the University of Bozen/Bolzano} in the academic year 202…
      • The goal of the course was to introduce the concept of Physics Informed Deep Neural Networks (PINN) and Neural Operators (NOs), discuss their implementation fro…
      • We discuss recent topics such as Mixture-of-Models, Fourier Neural Operators, Physics-Informed Kolmogorov-Arnold Networks (PIKANs) and Fourier Neural Operators.

  • Fathom: Per-Query Read Depth for Sparse Decoding over Offloaded KV Caches

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17652v1 Announce Type: new.
      • Abstract: When agentic sessions run to a million tokens with many sessions resident at once, the KV cache and the index that ranks it live in host memory, and the scan that ranks all n keys for a top-k step becomes the traffic that bounds decoding.
      • We present Fathom, a key scan in which each query decides how many bits of each key channel to read.
      • The 4-bit K cache is stored channel-major as bit planes, so a prefix of t planes is exactly the channel’s t-bit quantizer, and the query spends its bit budget by reverse water-filling over the variance-weighted importance of its channels.
    • EN 要点:
      • arXiv:2609.17652v1 Announce Type: new
      • Abstract: When agentic sessions run to a million tokens with many sessions resident at once, the KV cache and the index that ranks it live in host memory, and t…
      • We present Fathom, a key scan in which each query decides how many bits of each key channel to read
      • The 4-bit K cache is stored channel-major as bit planes, so a prefix of t planes is exactly the channel’s t-bit quantizer, and the query spends its bit budget b…

  • Reflect, Revise, Reuse: Training-Free Skill Evolution for GUI Agents

    • 发布时间:2026-09-17 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.17653v1 Announce Type: new.
      • Abstract: GUI agents execute long-horizon tasks on dynamic graphical user interfaces, where pop-ups, delayed loads, and relocated widgets routinely invalidate plans fixed before execution.
      • Recent agent-skill frameworks encapsulate reusable procedural knowledge to mitigate this, yet existing skill designs are largely developed without targeting GUI execution dynamics and treat skills as static artifacts produced before deployment rather than living procedural knowledge that improves through it.
      • We argue that what GUI agents need is not better static skills, but skills that can be revised from execution feedback at deployment time, without additional training.
    • EN 要点:
      • arXiv:2609.17653v1 Announce Type: new
      • Abstract: GUI agents execute long-horizon tasks on dynamic graphical user interfaces, where pop-ups, delayed loads, and relocated widgets routinely invalidate p…
      • Recent agent-skill frameworks encapsulate reusable procedural knowledge to mitigate this, yet existing skill designs are largely developed without targeting GUI…
      • We argue that what GUI agents need is not better static skills, but skills that can be revised from execution feedback at deployment time, without additional tr…