Andrej Karpathy, a co-founder of OpenAI, has proposed a new benchmark test that involves instructing an AI to generate a 3D world based on the opening of "The Lord of the Rings." He tested this concept with Claude Opus 5, providing it with the first paragraph of the book and a 1 million token budget. The AI produced approximately 5,500 lines of code over two hours, resulting in a 3D animation that recreates the scene of Bilbo Baggins' party. This benchmark is inspired by the "pelican on a bicycle" test, which evaluates an AI's ability to handle complex, creative instructions. AI
IMPACT This benchmark could drive advancements in AI's ability to handle complex, multi-modal creative tasks and code generation.
RANK_REASON Proposal of a new benchmark test for AI capabilities. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Mastodon — mastodon.social →
- Andrej Karpathy
- Claude 3.5 Sonnet
- Claude Opus 5
- ElevenLabs
- Google Gemini
- GPT-4o
- OpenAI
- The Lord of the Rings
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →