The Alignment Library

Comprehensive Knowledge Base on AI Alignment

A structured, comprehensive resource covering all fundamental problems, proposed solutions, and research frontiers in artificial intelligence alignment.

30+
Articles Techniques
6
Solutions Proposées
8
Problèmes Critiques

Core Problems

Explore fundamental challenges: outer alignment, inner alignment, corrigibility, and more.

Solutions & Research

Current approaches: RLHF, Constitutional AI, Debate, Mechanistic Interpretability, and their limitations.

Organizations & Researchers

Key players: MIRI, Anthropic, OpenAI, and leading researchers in the field.

Learning Resources

Curated reading lists, papers, videos, and courses organized by difficulty level.

A Note on P(doom)

This library presents alignment challenges honestly. Many leading researchers estimate very high probabilities of existential risk (50-99%+). The content reflects current technical understanding without false optimism.