AI Confidence Models
How CRADLE-0 presents uncertainty when recommendations are useful but disputed. This technical briefing turns a system concept into player-facing Mission Control language.
System Purpose
The system should expose a constraint that players can understand: threshold, deficit, redundancy, delay, maintenance, or confidence. It should explain how local work affects sector safety before it affects the planet.
Operational Tradeoff
Improvement always creates a second pressure. More capacity can add maintenance debt, more automation can reduce transparency, and more speed can export risk into ecology or governance.
UI And Event Use
Use this briefing for technology descriptions, blocked-state messages, event choices, and CRADLE-0 assessments. The language should stay concrete and auditable.
<!-- facility-package:start -->
Technology Facility Package
Technical Overview
AI confidence models make uncertainty visible before an instruction becomes policy. The facility is a public ledger of probability, evidence, and dissent.
Use Cases
Use it for CRADLE-0 advice, sensor disagreement, native signal interpretation, weather relay trust, and public override debate.
Failure Modes
Failure appears when a clean percentage hides sample bias, when low confidence is treated as ignorance, or when high confidence becomes permission to stop listening.
Linked Regions
- `world/echo-rift`
Mission Logs
- `logs/mission-log-014-ai-confidence`
- `logs/mission-log-013-biome-ledger`
- `logs/mission-log-011-weather-relay-audit`
Chronicle And Faction Dispute
- `chronicle/incident-ai-advisory-disclosure`
- `chronicle/cycle-070-biome-boundary-ledger`
- `chronicle/cycle-061-weather-relay-audit`
Demo Hook
The Demo hook is a confidence badge on risky actions: green does not mean moral clearance, and amber does not mean paralysis.
Visual Archive
Use public confidence interval art as the high-resolution plate for model uncertainty and review thresholds.
<!-- facility-package:end -->
