Скачать или смотреть ⚡️ARC-AGI-3: The Interactive Reasoning Benchmark

Greg Kamradt, President @ ARC Prize Foundation, joins us for a sneak preview of the ARC AGI 3 benchmark, how they are approaching game design for AI, and why this is a better benchmark than traditional knowledge-based ones. The preview challenge of the new benchmark has a $10,000 priz

Official website: https://three.arcprize.org/

X: https://x.com/latentspacepod

Substack: https://latent.space

Chapters

00:00:00 Introduction
00:00:47 Background on ARC AGI and the Benchmark
00:02:16 Defining Intelligence and Benchmark Philosophy
00:04:20 Challenges in Measuring Intelligence & Generalization
00:07:05 Introducing ARC AGI V3 and Interactive Games
00:08:26 Game Demo: Locksmith & Mechanics
00:12:30 Game Design, API, and Agent Competition
00:23:32 Team Structure and Building the Benchmark
00:24:43 Future Roadmap for ARC AGI Benchmarks
00:27:05 Durability, Timelines & Benchmark Evolution
00:29:17 The Role of ARC Prize in AGI Progress
00:32:05 Grok 4 Livestream & Model Evaluation
00:36:10 OpenAI, Competition, and the State of AI Labs
00:38:47 Closing Remarks & Call to Action