The Mechanics of Agency: Maximal Theoretical Agent
Summary
Explores Maximum Theoretical Agent (MTA)—opposite extreme from MVA. Agency spectrum from minimal (limited predictions, restricted influence) to maximal (comprehensive awareness of all timelines, complete control over branching).
Theoretical Foundations: Solomonoff Induction (predicting via algorithmic simplicity) + AIXI agent (Hutter’s optimal decision-maker using Solomonoff + reinforcement learning).
AIXI Demon Thought Experiment: Laplace’s Demon analogue for QBU. In “Quantum Garden” of future choices, AIXI Demon sees all branches with precise probabilities, perfect prediction of divergence/convergence including interference, optimal action selection steering timelines toward goals.
Example: Job decision abroad vs home—demon visualizes all timelines labeled by fulfillment/relationships/success, instantly determines optimal choice.
Implications: Clarifies agency spectrum, computational limits (MTA practically unattainable), ethical questions about responsibility/determinism, AI alignment insights (risks/benefits of powerful predictive systems).
Tags
Cross-References
- Related: Solomonoff Induction, AIXI
Notes
- MTA as theoretical ideal, not achievable goal
- Anticipates superintelligence discussions
- Quantum Garden metaphor recurs
- Alignment concerns introduced early