AI Is a Mirror

What outputs reveal about inputs.

"The mirror doesn't create. It reveals.", Kyle

Case ID: KR-001Active
← Back to Case Files
Case File
KR-001
AI Is a Mirror

AI reflects what we reward, ignore, and believe. The quality of the output reveals the quality of the input.

reflectionfeedbackcognitionbiassignalalignment
Inputs
Data
Prompts
Context
Incentives
Beliefs
Outputs
Answers
Insights
Decisions
Behaviors
Systems

Reflections, not origins

◈ Thesis

AI is a mirror of human systems. To get better outputs, improve the inputs, data, incentives, context, and clarity.

⚙ Mechanism
  • AI optimizes for patterns in training data and prompts.
  • Biases, gaps, and noise propagate to outputs.
  • Clarity, diversity, and alignment produce better reflections.
◎ Signals
  • Outputs feel generic or shallow.
  • Edge cases break reasoning.
  • Contradictions across similar queries.
  • Overconfidence in uncertain areas.
  • Drift over time without feedback.
▤ Notes
  • Garbage in, gospel out.
  • Prompts are part of the system, not just the interface.
  • Diversity in input > volume of input.
  • Feedback loops are essential.
  • The mirror can be polished, or distorted.
▥ References
  • The Alignment Problem, Brian Christian
  • Weapons of Math Destruction, Cathy O'Neil
  • The Filter Bubble, Eli Pariser
  • Superforecasting, Tetlock & Gardner
  • Poor Charlie's Almanack, C. Munger
? Open Questions
  • What am I rewarding without noticing?
  • Where is my input too narrow?
  • What would a cleaner mirror show me?