How Local Activists are Interrupting Big Tech AI Investments

• Sam Altman’s “AI empire” alone is using as much power as New York City and San Diego, combined
• Google’s AI data centers are consuming massive amounts of power, with electricity use hitting 30.8 million megawatt-hours in 2024, nearly doubling since 2020 — while inking deals with the Israeli military
• Elon Musk’s xAI has constructed the world’s largest AI supercenter, Colossus, in one of the poorest neighborhoods in one of the country’s most impoverished cities

Run LLMs on an AMD GPU. A cheaper alternative to running LLMs on an NVIDIA card?

The Enshittifinancial Crisis by Ed Zitron

Propose d’installer localement des commandes qui aident à «empower» un résolveur de mot-suivant.

This GitHub Gist provides Python code examples and results demonstrating a basic test of OpenAI’s Structured Output feature using the gpt-4o-mini model and pydantic to extract and structure data from image screenshots of financial disclosure reports and police blotter narratives.

Discover why OpenAI’s gpt-oss model family is ideal for building reliable and safe AI agents, not just for developers.

官网上对 Perplexity 的描述如下：

我在之前的文章《充分发挥 ChatGPT 在日常生活中的作用》中介绍过一款开源项目——ChatNextWeb，搭配 Vercel 即可部署自己的网页对话机器人，但它仅支持对话，如果想使用 DELL·E 3 模型进行文生图，似乎无法实现。最近看到另一个类似的项目——LobeCh

Altman has been reinstated 🤦🏻‍♀️😭💔

The Atlantic reports on how the for-profit rapid-commercialization camp’s leader was ousted and a hold-your-horses, we-need-to-only-use-AI-for-good guy was put in charge instead.

Sounds… good? One li’l worrying thing is that the people who leave are gonna bring their rah-rah-market-market ideology with them and unleash it.

Use GPT-3 incorrectly: reduce costs 40x and increase speed by 5x