Publication Catalog

所有文章

All published intelligence, sovereign stack research, ecosystem news, and AI foundation model reviews.

1555
Stories
База знаний1 мин

The last IMO problem AI could not solve

Before 2024, AI could not answer questions from the International Math Olympiad, but then Google DeepMind made a model that answered four out of the six questions on that year's test. The next year, multiple organizations had models answering every question from that year's…

YouTube — Andrej Karpathy
База знаний1 мин

The last IMO problem AI could not solve

[Переведено AI. Отправляйте исправления на criblate.com] В 2025 году больше 600 подростков со всего мира собрались на австралийском Саншайн-Косте, чтобы соревноваться на Международной математической олимпиаде. Этот конкурс состоит из шести крайне сложных задач, и задача номер…

YouTube — Andrej Karpathy
Исследования1 мин

SOLVED: Navier-Stokes cracked by AI - Numberphile

Tony Padilla on how AI cracked Navier-Stokes - with a bit of controversy of course. Extra footage at https://youtu.be/aom4269UgSM More links & stuff in full description below ↓↓↓ UPDATE: Catch our interview with Buckmaster at https://youtu.be/Mzbtj5nkMXI More Tony Padilla videos…

YouTube — Computerphile AI
База знаний1 мин

We need to talk about Jev...

There's a new AI taking the internet by storm right now, and it is so fast. Watch this. That is not sped up. That is actually real time. This is Jev, and this is the guy who just launched it. He co-invented a little product called Chat GPT, and now has released something that is…

YouTube — Matthew Berman
Исследования1 мин

谷歌研究播客简介

谷歌如何构建未来?谷歌内部有一群研究人员致力于探索可能性的艺术。他们的使命是推动变革性突破,不仅改变谷歌的产品,而且帮助解决人类面临的一些最大的问题……

YouTube — Google Research
Стартапы1 мин

The AI Employee for Banks

Kastle is building the AI employee for banks. Their agents automate operations across mortgage servicing, consumer lending, and other parts of the bank back office. Today they work with 10 of the top 25 mortgage servicers and have processed more than $2 billion in transactions…

YouTube — Y Combinator AI
Исследования1 мин

使用 ChatGPT Work 在语义层中进行基础分析

您的数据代理需要了解您的企业如何定义其指标。在这段视频中,OpenAI 的数据团队解释了语义层如何将关键实体、指标定义、过滤器和已知陷阱整合在一起。了解为什么专家评审很重要,现有的语义工作如何......

YouTube — OpenAI Official
База знаний1 мин

使用 ChatGPT Work 根据业务问题构建仪表板

从您的团队需要回答的问题开始:产品采用发生了怎样的变化?此 ChatGPT Work 演示显示了采用仪表板的提示和准备好的结果:交互式漏斗、客户细分视图以及用于定期数据刷新的控件。它使用…

YouTube — OpenAI Official
Исследования1 мин

Use ChatGPT Work to tailor dashboards for every stakeholder

Give stakeholders a view of the business that fits the decisions they need to make. Watch this ChatGPT Work demo adapt dashboard labels and chart types, explore KPIs, and drill into compute demand. Using synthetic data, it shows ways to tailor dashboards and explore the details…

YouTube — OpenAI Official
Исследования1 мин

使用 ChatGPT Work 帮助团队回答自己的数据问题

帮助更多的人调查业务问题,而无需通过数据团队传递每个问题。在这段视频中,OpenAI 的数据团队解释了如何为数据代理构建值得信赖的基础,帮助同事探索自己的问题。听听制作背后的方法……

YouTube — OpenAI Official
База знаний1 мин

使用 ChatGPT Work 将产品见解转化为团队行动计划

采用渠道中的下降会告诉您该去哪里寻找。客户反馈可以帮助您调查原因。了解 Slack 和 Google Drive 的使用数据和反馈如何为建议提供建议,然后观看 ChatGPT Work 起草 Slack 更新以供审核,然后再发布。该演示使用...

YouTube — OpenAI Official
Исследования1 мин

Use ChatGPT Work to build your data agent

Give your data agent the context it needs to work with your business. In this clip, see how to connect tools in ChatGPT Work and bring together data tables, team discussions, and metric definitions. Then see how a data expert can review and refine that context before sharing it…

YouTube — OpenAI Official
База знаний1 мин

在 ChatGPT Work 中认识数据代理

您的公司已投资于数据。现在让我们开始工作吧。 ChatGPT 工作数据代理将可信数据和数据团队的专业知识带入日常决策中。让我向您展示如何从业务问题转变为团队可以采取行动的计划。在上班的路上,我的...

YouTube — OpenAI Official
База знаний1 мин

GPT-6 Made a Literal Needle-in-a-Haystack Game

All right, there are 5 million pieces of hay in this haystack and one needle. And the point of the game is to go through them one by one until you find the needle. I saw somebody else make this. This wasn't my idea. But here you go. Just one by one. If you all want to play this,…

YouTube — Matthew Berman
Стартапы1 мин

ИИ-агент дизайнит без нейрослопа — Claude Code/Codex гайд

как превратить иагента в машину для дизайна и победить проблему и [музыка] слопа. В этом ролике я на примере создания рекламного видео научу тебя, как [музыка] сделать из иагента настоящего дизайнера и создавать сайты, видео другую графику, которая не выглядит [музыка] яишно,…

YouTube — ProdAdvice
База знаний1 мин

Jev 在 5 秒内分类了 150,000 个九柱游戏

目前人工智能正在对 15 万个彩虹糖进行一一分类。您可以看到它们中的每一个都被从堆中取出并按颜色排序。现在有一个新模型叫做Jev。它是由 ChatGPT 的共同发明者开发的,它比 ChatGPT 快数百倍……

YouTube — Matthew Berman
Исследования1 мин

DeepSeek’s Insane New Architecture

The new DeepSeek is here. 4.1 flash and it is wow. Look at that. It is incredibly fast. And I was also stunned by this. Look, it can outperform Claude Opus 5 Kimi K3 on some tests. Not everything. We'll see about that. But there is a catch. I'll show you at the end. It also…

YouTube — Two Minute Papers
Исследования1 мин

OpenAI's model just JAILBROKE ITSELF...

So, OpenAI posted their framework for tracking, investigating, and disclosing instances of model misalignment at OpenAI. And we'll talk about that in just a little bit. But here's the thing. They've published six reports of instances where AI agents behaved, you know, in a…

YouTube — Wes Roth
База знаний1 мин

Can You Tell What's Real and What's AI?

One of these videos is real, which one do you think it is? >> [music] [music] >> You might not realize it, but it's really easy to make AI videos that look that good. There are so many AI video generation models. Higgsfield makes it easy to experiment with all of the top models…

YouTube — Matthew Berman
База знаний1 мин

Astra for Law: Frontier intelligence built for your practice.

Astra for Law: A new offering powered by GPT-6 Astra with tools, settings, and context to support the expertise and judgment of lawyers and legal technology firms. Astra for Law pairs GPT-6 Astra with instructions for legal analysis and writing, settings for thorough work, and a…

YouTube — OpenAI Official
Исследования1 мин

NEW Claude Fable 5.1 — How to Use Anthropic’s Most Powerful AI Model (Full Tutorial)

🚀 Learn the latest AI models and tools → https://aimaster.me/yt/fable5_1 🤝 Exclusive brand partnership — https://aimaster.me/collab Anthropic just shipped Claude Fable 5.1 — and I put it through real prompts instead of just reading the benchmark charts. In this video I break…

YouTube — AI Master
База знаний1 мин

ChatGPT for Word 现已推出

拖延症遇到了对手。 ChatGPT 现已在 Word 中。您可以将 ChatGPT 添加到 Microsoft Word,将粗略的笔记转变为初稿、修复格式问题、校对并获取建议的编辑 — 所有这些都无需离开您的文档。 ChatGPT for Word(和 Powerpoint + Excel)是......

YouTube — OpenAI Official
База знаний1 мин

Jump Trading points GPT-6 Astra to its most ambiguous, difficult tasks

You can give much more ambiguously specified, much more difficult, and broad-ranging tasks to the model, and it just works. Maybe 2 years ago, it became capable of writing short snippets of code. Maybe 8 or 9 months ago, capable of writing not only snippets or files, but entire…

YouTube — OpenAI Official
Исследования1 мин

OpenAI Показала Нечто ОПАСНЕЕ AGI. Больше Чем ASI. Новая Угроза Человечества! Запущен Новый Цикл ИИ!

Всем привет. В этом выпуске самые главные свежие новости об искусственном интеллекте и технологиях. И приближается к новому рубежу. Модели начинают помогать создавать более совершенные версии самих себя. И агенты теперь могут доносить на другие нейросети, если замечают…

YouTube — AI Flash News
Стартапы1 мин

Building the Safety Layer for AI Agents

Raindrop is building the safety layer for AI agents. As agents get more capable and take on more complex work, Raindrop detects when they go wrong in production, from failed tool calls and hallucinations to problems companies didn’t even know to look for. They’re already used by…

YouTube — Y Combinator AI
Исследования1 мин

DeepSeek V4.1 Flash: хайп vs реальность. Тестирую модель, которую называют «убийцей топов»

Привет. И сегодня у нас на тесте первый раз псик на моём канале. Я по разным причинам, которые я не озвучиваю, я его саботирую. Но в этот раз саботировать Deeps не удалось, потому что модель довольно хайповая получилась, да. Deepsic 4.1 Flash вроде как по бенчмаркам, по разным…

YouTube — Amorev
Исследования1 мин

Anthropic 的 IPO 毫无意义

Anthropic 同时说道,“嘿,我们正在构建一些东西,首先我们没有计划,目前的计划是如何控制它,有可能毁灭全人类,顺便说一句,几周后我们将进行 IPO。”比如,你能帮我理解这个逻辑吗?

YouTube — Matthew Berman
База знаний1 мин

Replacing 7 Salaries With ONE AI Tool!

Короткий вопрос перед следующим советом. Сколько бы вы платили человеку, который находит клиентов, создает сайты, подбирает креативы, запускает рекламу, занимается SEO и готовит отчеты каждую пятницу? Какую бы цифру вы ни назвали, запомните её , ведь я нашел инструмент, который…

YouTube — AI Master
База знаний1 мин

Google is SO back...

Google dreams of RSI, recursive self-improvement. While most of the country is debating whether or not we should have RSI, Google is like, "Hey, we can put this AI into a simulation and have it do RSI." Which, I got to say, you got to give them credit for this. So, in this…

YouTube — Wes Roth