Import AI 476: Creativity and RSI; worse-than-China open weight models; AI goes to space
New benchmark exposes AI's visual reasoning gap—and worse-than-China open weights raise questions about the frontier labs' cost story.
Why it matters
Import AI 476 surfaces three substantive developments: a visual reasoning benchmark highlighting model shortcomings, open-weight models underperforming Chinese alternatives at lower cost, and early agentic AI deployments in space. Together, they complicate the narrative that frontier scale = quality and cost advantage.
The key facts
9 to knowHumanity's Sixth Sense benchmark demonstrates AI model gaps in intuitive visual reasoning
Open-weight models reported to underperform Chinese alternatives on cost-performance
Elorian and Scale AI mentioned as key players in AI research landscape
AI deployments reaching space applications
Article dated October 12, 2026 (Import AI #476)
Humanity's Sixth Sense benchmark highlights visual reasoning shortcomings in current AI systems
Scale AI and Elorian collaboration on benchmark work
Open-weight models underperforming relative to China-based alternatives
AI deployment expanding to space applications
Go to the source
Import AI (Blog)jack-clark.net
Publisher excerpt: Welcome to Import AI, a newsletter about AI research. Import AI runs on arXiv, cappuccinos, and feedback from readers. If you’d like to support this, please subscribe. Subscribe now New benchmark shows how AI models lack intuitive visual reasoning:…Humanity’s Sixth Sense highlights some…