1
2
搞定工业质检：AI视觉如何重构高端制造质量防线| 2026AI Partner·北京亦庄AI+产业大会
半导体光罩上的纳米级缺陷、玻璃基板上几微米的钻孔。广州因特智能展示了AI视觉如何从实验室走进高端制造产线，用软硬结合的方式解决中国在半导体检测装备领域的卡脖子问题。 广州因特智能孵化于 西安电子科技大学广州研究院 ，是校地合作落地的典型科技企业。我们拒绝“纸上算法”，坚持 软硬一体 ，为半导体、光通…
3
龙虾 OpenClaw 工程师示警：AI 正批量制造低质量危险代码
DeepSource 用 AI 自动审查代码漏洞与质量问题，帮团队避免低质量危险代码的批量产生
IT之家 5 月 23 日消息，华尔街日报昨日（5 月 22 日）发布博文，报道称参与打造“龙虾”OpenClaw 的 2 名工程师示警， AI 不只加快写代码速度，也可能把低质量代码批量扩散到真实产品与服务里。 OpenClaw 内部智能体框架 Pi 创建者马里奥 · 泽克纳（Mario Zech…
4
小米通报两起空调安装抽真空造假事件：涉事工程师永久拉黑
实时监控空调抽真空过程，数据云端上传确保安装质量闭环，提升服务透明度
IT之家 5 月 23 日消息，小米公司 5 月 19 日发布内部通报，针对在数字化抽真空服务专项稽核中发现的两起空调安装抽真空造假事件作出顶格处理。 根据网上流传出的公告截图，涉事的两名合作工程师被直接永久拉黑清退，终止任何形式的合作；其所属线下安装网点按虚假服务论处，每单罚款 1000 元，同时…
5
Quality and Security Signals in AI-Generated Python Refactoring Pull Requests
AI写代码重构PR靠不靠谱？这篇论文实测了质量与安全信号，开发者必读。
arXiv:2605.21453v1 Announce Type: cross Abstract: As AI agents increasingly contribute to code development and maintenance, there is still limited emp…
6
Tell HN: I'm tired of AI-generated answers
揭开AI答案的沉沦：开发者揭露GitHub讨论区充斥人工智能复读机，真实求助被淹没在无意义回复中。
I found GitHub repositories that were spreading malware. I asked AI what I should do about it, but it gave me nothing useful. So I opened a discussion…
7
Throughput vs. Goodput: The Performance Metricin LLM Testing
用卖Dosa的摊位类比，讲清LLM测试中常被忽略的吞吐量与有效吞吐量区别
Article URL: https://qainsights.com/throughput-vs-goodput-the-performance-metric-you-are-probably-ignoring-in-llm-testing/ Comments URL: https://news.…
8
Ask HN: Forbid Reddit HN Submissions?
Reddit充斥AI垃圾内容，HN社区热议是否该封杀其提交，引发对内容治理的深层思考。
cc @dang Reddit has been filling up with AI generated content. Either bots engaging with naïf audiences who don't realize they're bots, or supposed hu…
9
LLMs on the Line: Data Determines Loss-to-Loss Scaling Laws
ICML 2025收录，揭示数据质量如何决定大模型损失与缩放定律的深层关系。
arXiv:2502.12120v3 Announce Type: replace Abstract: Scaling laws guide the development of large language models (LLMs) by offering estimates for the o…
10
网易520发布会：质量为先，狙击细分赛道
5月20日，2026年『网易游戏520线上发布会』正式举办，公布了40余款游戏及IP的最新动态。其中，520发布会上公布的部分重要新品如下： 《遗忘之海》 《遗忘之海》 本次发布会上，《遗忘之海》官宣将于5月22日开启三测前瞻直播，并于28日正式开启测试。 目前来看，《遗忘之…
11
With Gemini 3.5 Flash, Google bets its next AI wave on agents, not chatbots
Google Gemini 3.5 Flash 驱动自主 AI 代理，低延迟高质量，专为编码和复杂任务执行而生
Google launched Gemini 3.5 Flash, its most powerful coding and agentic AI model yet, at the company's annual developer conference. It is capable of au…
12
Artifact-Bench: Evaluating MLLMs on Detecting and Assessing the Artifacts of AI-Generated Videos
新基准Artifact-Bench，专门评估多模态大模型识别AI视频伪影的能力，揭示模型真实短板。
arXiv:2605.18984v1 Announce Type: new Abstract: Recent video generative models have greatly improved the realism of AI-generated videos, yet their out…
13
Refreshed Nuxt ESLint Integrations
Nuxt迎来ESLint v9 flat config支持，新模块让代码集成更强大、更个性化。
We revamped our ESLint integrations to support ESLint v9 with the flat config, as well as a new module with many more capabilities.
14
Dual-Dimensional Consistency: Balancing Budget and Quality in Adaptive Inference-Time Scaling
LLM推理时缩放的新方法：双维度一致性，有效平衡采样预算与推理质量，提升效率。
arXiv:2605.15100v1 Announce Type: new Abstract: Large Language Models (LLMs) have demonstrated remarkable abilities in reasoning. However, maximizing …
15
Orchestrating AI Code Review at scale
Cloudflare分享大规模部署AI代码审查的实践经验，从瓶颈到解决方案，技术细节与工程思考兼具。
Learn about how we built a CI-native AI code reviewer using OpenCode that helps our engineers ship better, safer code.
16
Stitched Value Model for Diffusion Alignment
通过缝合价值函数实现扩散模型高效对齐，显著提升生成图像质量与一致性，为AI生成提供新范式
arXiv:2605.19804v1 Announce Type: cross Abstract: For practical use, diffusion- or flow-based generative models must be aligned with task-specific rew…
17
How Far Are We From True Auto-Research?
系统评估AI自动生成论文的质量差距，ResearchArena框架揭示可行性不等于真智能。
arXiv:2605.19156v1 Announce Type: cross Abstract: Recent auto-research systems can produce complete papers, but feasibility is not the same as quality…
18
Systematic Evaluation of the Quality of Synthetic Clinical Notes Rephrased by LLMs at Million-Note Scale
百万级临床笔记重写质量系统性评估，揭示LLM文本生成多维评价短板
arXiv:2605.17775v1 Announce Type: new Abstract: Large language models (LLMs) can generate or synthesize clinical text for a wide range of applications…
19
Critical Challenges and Guidelines in Evaluating Synthetic Tabular Data: A Systematic Review
系统性综述揭示合成表格健康数据评估的关键挑战与指南，为数据质量评估提供严谨框架
arXiv:2504.18544v3 Announce Type: replace-cross Abstract: Generating synthetic tabular health data is challenging, and evaluating their quality is equ…
20
How Much of the Internet Is AI Slop?
探讨互联网中AI生成的低质“垃圾内容”泛滥程度，以及它如何催化“脑腐”文化，拷问数字生态的未来。
Article URL: https://www.statsignificant.com/p/how-much-of-the-internet-is-ai-slop Comments URL: https://news.ycombinator.com/item?id=48207413 Points:…