Science ❯ Computer Science

AI Research

Model Evaluation Model Training Deep Learning Ethics in AI Model Development Machine Learning AI Development Human-Computer Interaction Generative Models AI Applications

Nvidia Launches Nemotron 3 Super, a 120B Open Model With 1M-Token Context for Agentic AI

Open weights with full training recipes aim to spur customization and scrutiny.

Studies Find AI Matches Average Creativity but Trails Top Humans, and Readers Penalize AI Labels

UC San Diego Puts Top AI Models Through Dungeons & Dragons, Finds Strengths—and Slippage Over Time

DeepMind’s Hassabis: China’s AI Models Are Only Months Behind U.S.

Yann LeCun Leaves Meta to Build AMI Labs on ‘World Models,’ Says LLMs Are a Dead End

Yann LeCun Says Meta "Fudged" Llama 4 Benchmarks as He Details Why He Left

OpenAI Releases GPT-5.2-Codex With Context Compression, Rolling Out to Paid ChatGPT

Yann LeCun Confirms AMI Labs, Names Alex LeBrun CEO as He Seeks €500 Million

Agentic RAG Gains Traction as Practitioners Map the Path to Production

Sergey Brin Calls His Retirement a Mistake, Rejoins to Drive Google’s Gemini

Stanford AI Agent Beats Most Human Pen Testers in Campus Network Trial

Google Launches Interactions API and Upgraded Gemini Deep Research for Developers

Musk Says AI Must Pursue Truth, Beauty and Curiosity to Stay Safe

Poetic Prompts Jailbreak AI Models in 62% of Tests, Researchers Report

Google’s Gemini 3 Accelerates With Benchmark Wins, Broad Rollout, and High‑Profile Praise

DeepMind Unveils SIMA 2, a Gemini-Powered Agent That Learns in 3D Game Worlds

Nature Publishes AlphaProof, DeepMind’s Verifiable Math AI Behind IMO Silver-Level Score

Deutsche Telekom and Nvidia Launch €1 Billion Industrial AI Cloud in Munich

King Charles Presses AI Safety With Nvidia’s Jensen Huang at QEPrize Ceremony

Stanford Study Finds Leading AI Models Fail to Distinguish Belief From Fact