The AI Fundamentalists Metaphysics and modern AI: What is causality?
Jan 27, 2026
They break down what makes a cause a cause and map common causal structures like chains, loops, and common causes. They contrast pure association with causal analysis and explain why randomized trials and natural experiments matter. They explore counterfactuals, first-principles models, and the statistical limits that shape trustworthy, causally grounded AI.
AI Snips
Chapters
Books
Transcript
Episode notes
Foundations Of Causal Relationships
- Causes and effects obey contiguity, precedence, and constant conjunction as philosophical constraints.
- Causal relationships take forms like chains, loops, shared causes, and shared effects which shape how we infer mechanisms.
Pattern Seeking Misleads Causal Judgement
- Correlation is distinct from causation and often misleads because humans impose narratives on patterns.
- Gestalt pattern-seeking makes us craft plausible causal stories from mere associations.
Use RCTs To Establish Causality
- Use randomized controlled trials (RCTs) to establish causal claims by randomly assigning treatment and control groups.
- Scale sample sizes to increase power and reduce the chance that effects are due to randomness.
