OpenAI Just Caught an AI Thinking! скачать в хорошем качестве

OpenAI Just Caught an AI Thinking!

OpenAI released circuit-sparsity, a research drop that exposes how a language model makes decisions internally. Instead of scaling up, OpenAI trained a transformer while cutting over 99.9% of its internal connections during training, forcing its logic into small, readable circuits. The release includes a real model and tooling that let researchers trace counting, memory, and decision-making step by step, arriving as OpenAI’s role in the AI economy grows more central and more sensitive to trust, control, and regulation. 👉 Join the waitlist for the twenty twenty-six AI Playbook https://tinyurl.com/AI-Playbook-2026 📩 Brand Deals and Partnerships: [email protected] ✉ General Inquiries: [email protected] 🧠 What You’ll See (Sources) • Weight-sparse transformers have interpretable circuits (paper) https://arxiv.org/abs/2511.13653 • OpenAI sparse circuits research overview (Official) https://openai.com/index/understandin... • openai/circuit-sparsity model on Hugging Face https://huggingface.co/openai/circuit... • openai/circuit_sparsity toolkit on GitHub https://github.com/openai/circuit_spa... • Axios article on OpenAI’s ecosystem impact https://www.axios.com/2025/12/13/open... 🚨 Why It Matters This isn’t about making AI smarter. It’s about making AI understandable. As models move deeper into code execution, content moderation, age gating, and real economic systems, internal decisions start to matter more than raw capability. Circuit-sparsity shows a path toward AI systems with fewer hidden interactions, traceable logic, and mechanisms humans can actually inspect.