Category: AI Safety

This category contains 16 pages.

• Corrigibility and the Off-Switch Problem
• Deceptive Alignment and Mesa-Optimization
• Goodhart's Law
• Instrumental Convergence
• Pascal's Mugging
• Specification Gaming and Reward Hacking
• The AI Box Experiment
• The Alignment Problem
• The Control Problem
• The Fragility of Value
• The Orthogonality Thesis
• The Paperclip Maximizer
• The Sharp Left Turn
• The Treacherous Turn
• The Waluigi Effect
• Wireheading