Scaling and the Road to Human-Level AI

Strategic framework for understanding AI scaling laws and building products that leverage predictable AI capability improvements.

Core Concepts

Two Phases of AI Training

Pretraining: Models learn to predict the next token by imitating human-written text, understanding underlying correlations in data.

Reinforcement Learning (RL): Models are optimized based on human feedback, reinforcing helpful/honest/harmless behaviors and discouraging harmful ones.

Scaling laws exist for both phases—performance improves predictably with increased compute, data, and parameters.

Key Metrics

• Task Horizon: Length/complexity of tasks AI can complete, measured in equivalent human time
• Elo Scores: Rating system measuring model preference comparisons
• Context Window: Amount of information processable in a single conversation