Chapter 6: Righteousness Growth and Performance Metrics

6.1 Righteousness Growth Score (RGS)
6.2 Tracking Righteousness Improvement Over Time
6.3 Righteousness Decision Metric (RDM)
6.4 Righteousness Performance Score (RPS)
6.5 Integrated RAGF Measurement Model

Part II — RAGF Measurement and Assessment System
Chapter 6: Righteousness Growth and Performance Metrics

6.1 Righteousness Growth Score (RGS)

The Righteousness Index (RI) provides a snapshot of AI righteousness at a single point in time. But righteousness is not a static state—it is a journey of continuous improvement. Organizations need to know not only where they stand today, but whether they are moving in the right direction over time.

The Righteousness Growth Score (RGS) answers this question. It measures the rate and magnitude of improvement in AI righteousness across the Five Pillars over a defined period.

Definition

RGS is a quantitative measure of the change in RI scores over time. It tracks whether an organization, AI system, developer practice, or agent is becoming more righteous—and at what pace.

RGS = RI(T2) − RI(T1)

Where:

RI(T1) = Righteousness Index score at the initial assessment
RI(T2) = Righteousness Index score at the subsequent assessment
The time interval between T1 and T2 is typically 6–12 months

What RGS Measures

RGS captures improvement across all five pillars:

This table 1 shows how the Righteousness Growth Score (RGS) is composed of pillar-level growth scores, each tracking improvement in a specific dimension of AI righteousness.

Table 1 — RGS Components

Component	What It Tracks	Interpretation
RGS-Integrity	Change in Integrity RI score	Is AI becoming more truthful and transparent?
RGS-Justice	Change in Justice RI score	Is AI becoming more fair and equitable?
RGS-Stewardship	Change in Stewardship RI score	Is governance and oversight improving?
RGS-Wisdom	Change in Wisdom RI score	Is decision quality and foresight improving?
RGS-Beneficence	Change in Beneficence RI score	Is AI contributing more to human flourishing?
RGS-Overall	Change in overall RI score	Is overall AI righteousness improving?

This table 2 provides guidance on interpreting RGS scores and the recommended actions for each score range.

Table 2 — RGS Score Interpretation

RGS Range	Interpretation	Recommended Action
+10 or more	Significant improvement	Maintain momentum; identify and scale successful practices
+5 to +9	Moderate improvement	Continue current trajectory; address any remaining gaps
0 to +4	Minimal improvement	Investigate barriers; reassess implementation effectiveness
−1 to −5	Slight decline	Conduct root cause analysis; identify negative drivers
−6 or less	Significant decline	Urgent intervention required; comprehensive governance review

When to Measure RGS

Assessment Type	Timing	Purpose
Baseline Assessment	Initial	Establish starting point
Follow-up Assessment	6 months	Early progress check
Annual Assessment	12 months	Full growth evaluation
Periodic Reviews	Quarterly	Monitor ongoing trends

RGS in Practice

RGS provides organizations with a clear, actionable indicator of whether their righteousness governance efforts are delivering results. A positive RGS validates that investments in AI governance are yielding measurable improvements. A negative RGS signals the need for course correction before problems become entrenched.

Key Insight: RGS is not about achieving perfection—it is about progress. Even small positive growth over time compounds into significant long-term improvement. The goal is not just to be righteous, but to be growing in righteousness.

6.2 Tracking Righteousness Improvement Over Time

Sustained righteousness requires systematic tracking of improvement over time. RGS provides the quantitative foundation, but organizations need a broader framework for monitoring, analyzing, and acting on righteousness trends.

The Righteousness Growth Cycle

Assessment → Analysis → Action → Re-assessment → Growth

This cycle ensures that righteousness governance is not a one-time effort but a continuous process of learning and improvement.

Key Tracking Mechanisms

Table 3 — Righteousness Tracking Mechanisms

Mechanism	Purpose	Frequency
RI Re-assessment	Measure current righteousness status	Annually or semi-annually
RGS Calculation	Quantify improvement over time	After each re-assessment
Trend Analysis	Identify patterns and trajectories	Quarterly
Pillar-Level Review	Assess progress on specific dimensions	Quarterly
Incident Tracking	Monitor righteousness failures	Continuous
Stakeholder Feedback	Gather qualitative insights	Ongoing

Trend Analysis

Trend analysis examines the direction and consistency of righteousness scores over multiple assessment periods. It helps identify:

Trend Pattern	Implication
Consistent upward trend	Governance efforts are effective; continue current approach
Inconsistent or volatile	Governance is unstable; identify sources of variability
Plateau	Improvement has stalled; need new interventions
Downward trend	Governance is deteriorating; urgent intervention required

Pillar-Level Growth Monitoring

Not all pillars improve at the same rate. Tracking growth at the pillar level helps organizations identify:

Which dimensions are responding well to governance efforts
Which dimensions are lagging and need additional attention
Whether improvements in one pillar are correlated with improvements in others

Dashboard for Righteousness Tracking

A righteousness tracking dashboard provides real-time visibility into:

Current RI scores for all assessed entities
RGS trends over time
Pillar-level breakdowns
Alerts for significant changes or declines
Comparison to benchmarks or prior periods

Key Insight: Regular tracking transforms righteousness from an abstract aspiration into a managed, measurable process. Organizations that track righteousness systematically are better positioned to sustain and accelerate their growth.

6.3 Righteousness Decision Metric (RDM)

The Righteousness Index (RI) measures the overall state of AI righteousness. The Righteousness Growth Score (RGS) measures improvement over time. But organizations also need to evaluate the righteousness of individual decisions made by AI systems.

The Righteousness Decision Metric (RDM) addresses this need. It evaluates whether a specific AI decision—or a set of decisions—is righteous, based on the Five Pillars.

Definition

RDM is a metric that assesses the righteousness of individual AI decisions or decision patterns. It evaluates decisions against the Five Pillars to determine whether they align with righteousness principles.

RDM Assessment Dimensions

This table 4 outlines the dimensions used to assess the righteousness of individual AI decisions through the Righteousness Decision Metric (RDM).

Table 4 — RDM Assessment Dimensions

Dimension	Assessment Question	Indicator
Integrity	Was the decision truthful and transparent?	Decision is based on accurate information; limitations are disclosed
Justice	Was the decision fair and non-discriminatory?	Decision treats all affected parties equitably
Stewardship	Was the decision made with appropriate oversight?	Decision was reviewed or authorized as required
Wisdom	Was the decision prudent and well-reasoned?	Decision considers consequences and alternatives
Beneficence	Does the decision promote human well-being?	Decision contributes to positive outcomes

RDM Calculation

RDM is typically calculated as a score (0–100) for each decision or as an aggregate score for a set of decisions:

RDM = Average of scores across the Five Pillars for a given decision

This table 5 provides guidance on interpreting RDM scores and the recommended actions for each score range.

Table 5 — RDM Score Interpretation

RDM Range	Interpretation	Recommended Action
80–100	Righteous decision	Document as best practice; use as training example
60–79	Generally righteous	Minor improvements possible; review for optimization
40–59	Mixed righteousness	Significant concerns; review decision process
20–39	Questionable righteousness	Immediate review required; consider decision reversal
0–19	Unrighteous decision	Urgent intervention; reverse decision if possible

RDM in Practice

RDM is particularly valuable for:

Application	Description
Real-time Decision Monitoring	Evaluating AI decisions as they are made
Audit and Compliance	Reviewing past decisions for righteousness
Training and Improvement	Identifying patterns of unrighteous decisions
Risk Management	Flagging high-risk decisions for human review

Key Insight: RDM provides granular, decision-level insight that complements the broader RI and RGS metrics. While RI tells you the overall state and RGS tells you the trajectory, RDM tells you whether specific decisions are righteous.

6.4 Righteousness Performance Score (RPS)

The Righteousness Performance Score (RPS) evaluates the effectiveness of governance activities in producing righteous outcomes. While RI measures the state of righteousness and RGS measures growth, RPS measures how well the governance system itself is performing.

Definition

RPS is a metric that assesses the performance of AI governance activities—policies, processes, controls, and interventions—in achieving righteous outcomes.

RPS Assessment Dimensions

This table 6 outlines the dimensions used to assess governance performance through the Righteousness Performance Score (RPS).

Table 6 — RPS Assessment Dimensions

Dimension	Assessment Question	Indicator
Coverage	Are all AI systems and activities governed?	Percentage of AI systems covered by governance
Effectiveness	Are governance activities achieving their objectives?	RI improvement correlated with governance activities
Efficiency	Are governance activities cost-effective?	Resources required per unit of improvement
Responsiveness	Does governance adapt to emerging issues?	Time to address identified gaps
Maturity	Is governance becoming more sophisticated over time?	Governance capability scores

This table 7 provides guidance on interpreting RPS scores and the recommended actions for each score range.

Table 7 — RPS Score Interpretation

RPS Range	Interpretation	Recommended Action
80–100	Excellent governance performance	Maintain and scale best practices
60–79	Good governance performance	Identify optimization opportunities
40–59	Adequate governance performance	Significant improvements needed
20–39	Poor governance performance	Comprehensive governance redesign
0–19	Ineffective governance	Urgent governance overhaul required

RPS in Practice

RPS is valuable for:

Application	Description
Governance Audit	Assessing the effectiveness of governance structures
Resource Allocation	Identifying where additional resources are needed
Continuous Improvement	Tracking governance maturity over time
Stakeholder Reporting	Demonstrating governance effectiveness to stakeholders

Key Insight: RPS answers the question: “Are our governance efforts actually working?” Good RI scores may reflect strong governance—or they may reflect luck. RPS helps distinguish between the two by evaluating the systems and processes that produce righteous outcomes.

6.5 Integrated RAGF Measurement Model

The RAGF Measurement Model integrates RI, RGS, RDM, and RPS into a coherent, multi-dimensional system for measuring AI righteousness at all levels.

The Four Metrics

This table 8 provides an overview of the four core RAGF metrics—RI, RGS, RDM, and RPS—including their focus, question, and use.

Table 8 — The Four RAGF Metrics

Metric	Focus	Question	Primary Use
RI (Righteousness Index)	Current state	“How righteous are we now?”	Baseline assessment, benchmarking, certification
RGS (Righteousness Growth Score)	Improvement over time	“Are we becoming more righteous?”	Progress tracking, trend analysis, motivation
RDM (Righteousness Decision Metric)	Individual decisions	“Is this decision righteous?”	Real-time monitoring, audit, training
RPS (Righteousness Performance Score)	Governance effectiveness	“Are our governance efforts working?”	Governance audit, resource allocation, improvement

How the Metrics Work Together

Figure 1: Integrated RAGF Measurement Model

This figure 1 illustrates how RI, RGS, RDM, and RPS work together as an integrated measurement system, with RPS assessing governance performance, RI measuring current state, RGS tracking growth, and RDM providing decision-level insight.

This table 9 shows how the four RAGF metrics relate to and depend on each other.

Table 9 — Metric Relationships and Dependencies

Primary Metric	Dependent On	Relationship
RI	Assessment data, evidence collection	Provides the foundational measurement
RGS	RI at T1 and T2	Calculated from changes in RI over time
RDM	Decision logs, real-time monitoring	Provides granular insight below the RI level
RPS	RI, RGS, and governance activity data	Evaluates whether governance is driving improvement

Use Cases for the Integrated Model

Use Case	Metrics Used	Application
Certification	RI	Assessing whether an AI system meets righteousness standards
Progress Reporting	RI + RGS	Demonstrating improvement over time
Governance Audit	RPS + RI	Evaluating governance effectiveness
Incident Investigation	RDM + RI	Analyzing specific decisions in context
Strategic Planning	All four	Informing governance strategy and resource allocation

References

Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human behavior. Plenum Press.

Dignum, V. (2019). Responsible artificial intelligence: How to develop and use AI in a responsible way. Springer.

Floridi, L., & Cowls, J. (2019). A unified framework of five principles for AI in society. Harvard Data Science Review, 1(1).

Floridi, L., Cowls, J., Beltrametti, M., Chatila, R., Chazerand, P., Dignum, V., … & Vayena, E. (2018). AI4People—An ethical framework for a good AI society: Opportunities, risks, principles, and recommendations. Minds and Machines, 28(4), 689–707.

Kuang, Q. (2026). Deed & Ethics Quotient Model (DEQM): A universal quantitative system for virtue measurement across humans, AI, and intelligent machines. Journal of Ethical AI Research, 3(1).

Metlushko, S. (2026). A formal deontic framework for AI decision evaluation: The Kant Code. PhilPapers.

National Institute of Standards and Technology. (2023). Artificial Intelligence Risk Management Framework (AI RMF 1.0). U.S. Department of Commerce.

OECD. (2019). OECD Principles on Artificial Intelligence. OECD Publishing.

Ryan, R. M., & Deci, E. L. (2017). Self-determination theory: Basic psychological needs in motivation, development, and wellness. Guilford Press.

Smuha, N. A. (2021). From a race to the bottom to a race to the top? The regulation of AI and the EU AI Act. European Journal of Legal Studies, 13(2), 45–72.

Walther, C. (2026). Is your AI system ethical? Try this assessment. Knowledge@Wharton.

Wang, K., Wang, Z., Lv, P., Li, H., Li, G., Cong, G., & Dong, W. (2026). Data agents under attack: Vulnerabilities in LLM-driven analytical systems. arXiv preprint arXiv:2606.08661.