朝花夕拾

Daily Vault

8月3日(月)（20件）

Apple is getting this wrong

OpenAI addresses Apple’s baseless lawsuit, corrects claims about its employees, and shares messages documenting what happened.

·AI/, ·openai.com ↗

openai.com ↗

搜索引擎公司 DuckDuckGo“整活”推出 Paso Robles 太阳镜，号称无 AI、无摄像头、永不发送数据

IT之家 8 月 3 日消息，谷歌、三星、Meta 等科技巨头正在争相布局 AI 智能眼镜市场，镜框内置的微型摄像头，则逐渐让智能眼镜成为公众的隐私“噩梦”，利用其导致的偷拍和其他滥用事件呈增加趋势。注重隐私的搜索引擎公司 DuckDuckGo 上月末（7 月 30 日）宣布推出 Paso Robles 太阳眼镜，宣传语是“普通得不能再普通的太阳镜”（Normal F*ing Sunglasses）。据介绍，这款太阳镜联合 Kn

·AI/, ·ithome.com ↗

ithome.com ↗

中国信通院：2025 年我国人工智能产业规模超 1.2 万亿元，同比增长 40%

IT之家 8 月 3 日消息，中国信息通信研究院今日发文，经测算，2025 年我国人工智能产业规模超 1.2 万亿元，同比增长 40% 。 IT之家获悉，截至 2026 年 6 月，人工智能企业数量超 6600 家，全球占比达到 15% ，已形成覆盖基础底座、模型框架、行业应用的完整产业体系。人工智能产业链分为基础层、模型框架层和应用层。2025 年，我国人工智能基础层产业规模同比增长 59%，占比为 38%；人工智能模型框架层规

·AI/·ithome.com ↗

ithome.com ↗

中国信通院：2025年我国人工智能产业规模超1.2万亿元

36氪获悉，中国信息通信研究院持续开展人工智能产业研究，并开展2025年人工智能产业规模测算。经测算，2025年我国人工智能产业规模超1.2万亿元，同比增长40%。截至2026年6月，人工智能企业数量超6600家，全球占比达到15%，已形成覆盖基础底座、模型框架、行业应用的完整产业体系。

·AI/·36kr.com ↗

36kr.com ↗

Qwen3.8-Max上线国家超算互联网

36氪获悉，8月3日，阿里巴巴正式发布新一代基座大模型Qwen3.8-Max，总参数量达到2.4万亿，成为继KimiK3之后又一突破两万亿参数的国产大模型。发布当日，国家超算互联网上线该模型API调用服务，开发者可通过官网直接接入。

·AI/·36kr.com ↗

36kr.com ↗

At colleges, the AI boom means everyone wants to dabble in computer science

With no background in coding, Faith Maeba, a psychology major, was reluctant when her mother first suggested she enroll in classes on artificial intelligence.

·AI/, ·phys.org ↗

phys.org ↗

自 2023 年，谷歌、亚马逊、微软等美国科技巨头已在 AI 领域砸下 1 万亿美元

IT之家 8 月 3 日消息，据《金融时报》前天报道，自 2023 年 AI 热潮兴起以来，谷歌、亚马逊、微软和 Meta 四家美国科技巨头已在 AI 领域投入 1.1 万亿美元（IT之家注：现汇率约合 7.44 万亿元人民币）。这其中的大部分资金用于数据中心、芯片和电力基础设施建设。据报道，这笔巨额的 AI 投资已经影响供应链以及苹果等竞争对手，且上述科技巨头的 AI 支出暂时没有任何放缓迹象，这四家公司今年的 7450 亿美

·AI/, ·ithome.com ↗

ithome.com ↗

How we built a realtime system for responsive voice AI in six months

GPT-Live enables continuous voice interaction with AI, using a turnless speech model and low-latency architecture for faster, more natural conversations.

·AI/, ·openai.com ↗

openai.com ↗

刚刚，阿里Qwen3.8-Max来了！冲进全球第一梯队，模型表现直逼Claude

编程、专业工作、长程任务、多模态分析统统梭哈

·AI/, ·qbitai.com ↗

qbitai.com ↗

ConnectED: A Curriculum-Aligned AI System for Vietnamese Instructional Lesson Planning and Student Learning

arXiv:2607.28647v1 Announce Type: cross Abstract: This paper presents ConnectED, a human-centered AI system that supports the full instructional lifecycle in Vietnamese education by linking curriculum-aligned lesson desi

·AI/·arxiv.org ↗

arxiv.org ↗

The Formalism Trap: Are LLM-as-a-Judge Evaluators Blinded by Consensus Mimicry under Social Load?

arXiv:2607.28641v1 Announce Type: cross Abstract: We introduce the and the Evaluative Dissonance Index ($D_E$), quantifying how LLM-as-a-Judge systems conflate structural proceduralism with semantic truth under adversari

·AI/·arxiv.org ↗

arxiv.org ↗

Topology-Aware Data Movement for Disaggregated GPU Inference

arXiv:2607.28633v1 Announce Type: cross Abstract: Disaggregated LLM inference creates a datacenter networking problem that no existing system solves correctly. When prefill and decode run on separate GPU pools, the KV ca

·AI/, , ·arxiv.org ↗

arxiv.org ↗

Scaffolding Critical Engagement with GenAI: Transforming Ethnic Minority Preparatory Students' Collaborative Discourse in Prompt Engineering Tasks

arXiv:2607.28630v1 Announce Type: cross Abstract: Generative AI (GenAI) holds significant promise for advancing educational equity among ethnic minority students by broadening access to learning resources and mitigating

·AI/·arxiv.org ↗

arxiv.org ↗

AgentHPOBench: A Benchmark For Evaluating LLM Agents as Sequential Hyperparameter Optimizers

arXiv:2607.29626v1 Announce Type: new Abstract: As LLMs evolve from code completion systems into autonomous scientific agents, evaluating their ability to conduct experiments has become increasingly important. Existing b

·AI/, ·arxiv.org ↗

arxiv.org ↗

LEMUR: Learning to Align with Multi-Objective Reinforcement Learning from Preference Feedback

arXiv:2607.29559v1 Announce Type: new Abstract: Reinforcement Learning (RL) systems are typically trained using a single, well-specified scalar reward function. However, real-world decision-making tasks often involve mul

·AI/·arxiv.org ↗

arxiv.org ↗

AMTFV: Agentic Mathematical Tool-Flow Verification for LLM Self-Correction

arXiv:2607.29549v1 Announce Type: new Abstract: Large language models have demonstrated strong mathematical problem-solving capabilities, yet reliably verifying their candidate answers remains challenging. Existing repre

·AI/, ·arxiv.org ↗

arxiv.org ↗

ModelEquivBench: Certifying Multi-Relational Evaluation of LLM-Generated Optimization Models

arXiv:2607.29431v1 Announce Type: new Abstract: Large language models increasingly generate optimization models from natural language, but existing evaluation often reduces a generated model and its ground truth to a sin

·AI/·arxiv.org ↗

arxiv.org ↗

Beyond Component Testing: Validating Agentic AI Systems

arXiv:2607.29405v1 Announce Type: new Abstract: Agentic AI systems act through multi-step trajectories that combine planning, tool use, memory, interaction, and adaptation. This behavior stretches validation practice bey

·AI/, , ·arxiv.org ↗

arxiv.org ↗

Tool Specifications Matter: Uncovering and Mitigating Safety Risks in AI Agents

arXiv:2607.29254v1 Announce Type: new Abstract: AI agents extend large language models (LLMs) with external tools, enabling them to perform complex tasks and translate model outputs into consequential real-world actions.

·AI/, , ·arxiv.org ↗

arxiv.org ↗

MirrorCraft: Paired Evaluation under Hidden Rule Changes in Minecraft

arXiv:2607.29218v1 Announce Type: new Abstract: With the prosperity of the large language models (LLMs), it has become an interesting topic: how do LLM-based agents work in Minecraft? Unfortunately, most existing benchma

·AI/, ·arxiv.org ↗

arxiv.org ↗