This is a place where I think in public on AI dangers

Concepts & Language

This is pre-reading, aimed at readers who are both non-technical, and/or not familiar with how today's LLMs work

• Levels of AI 'Aliveness'

• Technical Concepts

Articles

• Logistics of a rogue agent attack

• Runaway AI: Potentially Accessible Compute and Inference Capacity

Concepts

• Accidental Rogue Agent

• Extreme Survival and Replication Instincts

• Vulnerable Surfaces

  • Vulnerable Surface Compute Capacity