66 stories
Una evaluación piloto del riesgo de despliegue no autorizado en empresas de IA de frontera. En febrero de 2026, METR inició un ejercicio piloto para evaluar riesgos de desalineación derivados de agentes de IA usados dentro de empresas desarrolladoras de IA de frontera, con la participación de…
2026 年 2 月，METR 开展了一项试点研究，对前沿 AI 公司在实验室内部使用 AI 智能体时可能出现的不对齐风险进行了评估。Anthropic、Google、Meta 和 OpenAI 参与了这项试点。报告正文分为三部分：
A pilot assessment of rogue deployment risk at frontier AI companies. Starting in February 2026, METR conducted a pilot exercise to assess misalignment risks from AI agents used inside frontier AI developers, with participation from Anthropic, Google, Meta, and OpenAI.
127 summer olympics trivia questions with answers covering Paris 2024 records, Phelps, Bolt, Biles, ancient Greece, and the LA 2028 preview.
The chip inside your smartphone is the product of one of the most precise manufacturing processes ever devised by humanity. To build it, engineers must draw patterns smaller than a virus onto silicon
A NIST researcher has unveiled the results of a 10-year quest to measure the constant.
Special thanks to John Schulman for a lot of super valuable feedback and direct edits on this post. Test time compute (Graves et al. 2016, Ling, et al. 2017, Cobbe et al. 2021) and Chain-of-thought (CoT) (Wei et al. 2022, Nye et al.
Posted by Dustin Zelle, Software Engineer, Google Research, and Arno Eigenwillig, Software Engineer, CoreML Objects and their relationships are ubi...
Announcing the release of TensorFlow GNN 1.0, a production-tested library for building GNNs at Google scale, supporting both modeling and training.
The use of large language models in the real world has strongly accelerated by the launch of ChatGPT. We (including my team at OpenAI, shoutout to them) have invested a lot of effort to build default safe behavior into the model during the alignment process (e.g. via RLHF).
[Updated on 2021-02-01: Updated to version 2.0 with several work added and many typos fixed.] [Updated on 2021-05-26: Add P-tuning and Prompt Tuning in the “prompt design” section.] [Updated on 2021-09-19: Add “unlikelihood training”.]
[Updated on 2020-11-12: add an example on closed-book factual QA using OpenAI API (beta). A model that can answer any question with regard to factual knowledge can lead to many useful and practical applications, such as working as a chatbot or an AI assistant🤖.
Showing 66 of 66 stories
You're all caught up.
Get the top AI news and tools straight to your inbox every week.
Weekly · AI news
Join 15,000+ AI enthusiasts
