About AI Daily Radar

AI Daily Radar is a free daily briefing on artificial intelligence and the frontier technology around it. Every morning (US Eastern Time) we publish one issue in English and the same issue in Chinese.

What we cover

Seven sections, in this order: AI, Chips & Computing, Robots & Self-Driving, EVs & Energy (Tesla included), Space, Science & Biotech, and Tech Industry & Policy, plus a few Deep Dives from analysis newsletters.

What we read

Around 400 to 500 new items a day from more than 60 sources:

AI labs: OpenAI, Google DeepMind, Google Research, Microsoft Research, Hugging Face, Berkeley AI Research, NVIDIA.
Technology press: Techmeme, Ars Technica, MIT Technology Review, IEEE Spectrum, Wired, The Verge, TechCrunch, Rest of World, The Decoder.
Chips and computing: SemiAnalysis, Tom’s Hardware, ServeTheHome, EE Times, The Quantum Insider.
Robots, EVs and energy: The Robot Report, Electrek, Teslarati, Not a Tesla App, InsideEVs, Electrive, Canary Media, pv magazine, World Nuclear News.
Space and science: SpaceNews, NASASpaceflight, Space.com, Quanta Magazine, STAT, MIT News.
Communities and search: Hacker News; Reddit communities on machine learning, hardware, robotics, Tesla, EVs, energy and space; GitHub releases of major AI projects; Google News searches for AI, chips, robots, Tesla and energy.
Chinese tech media: QbitAI (量子位), AI Era (新智元) and GeekPark (极客公园), so readers also see what is moving in China.

How stories are picked

A large language model sorts each item into a section and scores it from 0 to 10 against that section’s written rubric: is it really about this field, is it new, who is affected, how credible is the source. Rumours and stock chatter are marked down; stories that fit no section are dropped. Duplicate reports of the same event are merged, and every section and source type has a cap so no single topic fills the issue. At most 40 stories make it in.

How summaries are written

For each selected story, a language model writes a short summary, background and, where the evidence supports it, the practical impact. It is instructed to separate announced plans from shipped products, attribute opinions and avoid hype. Every story links to its original source, and the score shown beside each headline is the model’s rating.

Because the summaries are machine-written, they can contain mistakes. Always check the original source before relying on a detail.

Corrections and contact

Spotted an error, or want to suggest a source? Email hello@aidailyradar.com. We fix confirmed errors and adjust the rules that caused them.

Advertising

The site is funded by advertising served by Google AdSense. Advertisers have no say in which stories are selected or how they are described. See the privacy policy for how ads use cookies.

关于 AI 每日雷达

AI 每日雷达是一份关于人工智能及相关前沿科技的免费每日简报。每天早上（美东时间）发布一期，同时提供英文版和中文版。

报道范围

分七个板块，依次是：人工智能、芯片与算力、机器人与自动驾驶、电动车与新能源（包括特斯拉）、航天、科学与生物科技、科技产业与政策，另有少量来自分析通讯的深度解读。

我们读什么

每天大约 400 到 500 条新内容，来自 60 多个信息源：

AI 实验室： OpenAI、Google DeepMind、Google Research、Microsoft Research、Hugging Face、伯克利 AI 研究院、NVIDIA。
科技媒体： Techmeme、Ars Technica、MIT Technology Review、IEEE Spectrum、Wired、The Verge、TechCrunch、Rest of World、The Decoder。
芯片与算力： SemiAnalysis、Tom’s Hardware、ServeTheHome、EE Times、The Quantum Insider。
机器人、电动车与能源： The Robot Report、Electrek、Teslarati、Not a Tesla App、InsideEVs、Electrive、Canary Media、pv magazine、World Nuclear News。
航天与科学： SpaceNews、NASASpaceflight、Space.com、Quanta Magazine、STAT、MIT News。
社区与搜索： Hacker News；Reddit 上机器学习、硬件、机器人、特斯拉、电动车、能源和航天社区；主要 AI 开源项目在 GitHub 上的版本发布；Google 新闻中关于 AI、芯片、机器人、特斯拉和能源的搜索结果。
中文科技媒体： 量子位、新智元、极客公园，让读者同时看到中国科技行业的动向。

怎么选新闻

大语言模型先把每条内容分到对应板块，再按该板块成文的评分标准打 0 到 10 分：是否真属于这个领域、是不是新消息、影响谁、来源是否可靠。传闻和股价闲聊会被降分，哪个板块都不属于的内容直接剔除。同一事件的多篇报道会合并；每个板块、每类来源都有上限，避免单一话题占满一期。每期最多收录 40 条。

摘要怎么写

每条入选新闻由大语言模型撰写简短摘要、背景，以及有证据支撑时的实际影响。写作规则要求区分”宣布的计划”和”已经上线的产品”，标明观点出处，不夸大。每条都附原始来源链接，标题旁的分数是模型的评分。

摘要由机器生成，可能有错误。引用具体细节前，请先核对原文。

更正与联系

发现错误，或想推荐信息源？请发邮件到 hello@aidailyradar.com。确认的错误我们会更正，并调整导致错误的规则。