Dimension B: AI as the Witnessing Model’s Scale Engine
What It Does
Makes human witnessing operationally viable at national scale without compromising the integrity that makes it valuable. The witnessing model’s integrity rests on human confirmation anchoring records in physical reality. That principle cannot change. But Leja cannot have a human reviewer touching every record at scale. AI’s role: carry the weight everywhere human judgment is not load-bearing, so human attention is concentrated exactly where it matters.The Division of Labor
AI HANDLES:- Document pre-screening: flags anomalies before human review
- Consistency checking: cross-references claimed information against uploaded evidence and existing records
- Submission triage: identifies the 3% of submissions that need human attention so reviewers touch 3%, not 100%
- Regional pattern learning: what legitimate Nigerian tenancy documents look like across 36 states — regional variation does not create false flags
- Fraud pattern surfacing: emerging network-level patterns presented to human reviewers before individual cases escalate
- Maintenance request categorization: routes to correct service category
- Consequential attestations: fraud flags, dispute resolutions, formal record challenges
- Edge cases: anything AI flags as outside its confidence threshold
- Appeals: any party disputing an AI-assisted decision
- Policy interpretation: novel situations not covered by existing rules