TIS
Truth Integrity Score surfaces reliability, consistency, uncertainty, contradiction, and missing context rather than presenting confidence as certainty.
Architecture
Supervision is a system boundary, not an afterthought.
EthosGate separates an agent's ability to propose from its authority to act. The evaluation path makes ethical constraints and uncertainty available for review before consequential execution proceeds.
Decision path
Propose, evaluate, decide, record.
The implementation remains intentionally modular so truth assessment, autonomy assessment, societal impact, policies, and audit behavior can each be examined independently.
01
Context ingestion
Gather the proposed action, objective, available evidence, material unknowns, and the scope of potential impact.
02
Conscience layer
Evaluate truth integrity, human autonomy, and societal resonance before a downstream system is authorized.
03
Decision engine
Compare the request against constitutional invariants, applicable policy overlays, uncertainty, consequence, and reversibility.
04
Human authority and audit
Keep review explicit when needed and append the decision context, authority, and outcome to an accountable record.
HAI
Human Autonomy Index evaluates whether the proposed action preserves meaningful choice, dignity, and the ability to override the system.
SRQ
Societal Resonance Quotient considers harm, fairness, third parties, downstream consequences, and wider systemic effects.
Constitutional invariants remain distinct from organization policy.
Policies may adapt to an organization's context. Constitutional safeguards establish the non-negotiable boundary: protect human life, dignity, autonomy, truthfulness, and accountability; do not knowingly create unjustified harm.
View corpus provenance