Cognition · Software component

Utility-Based Decision Maker

Software componentCognitionCognition & Memoryarc:UtilityBasedDecisionEngine

A decision engine that scores each candidate action by expected utility, the probability-weighted sum of outcome utilities, and selects the action maximizing it.

Responsibility. Selects the action with maximum expected utility under a specified utility function.

Also known as: Utility-based agent, MEU agent, Expected-utility maximizer, Scalarized multi-objective decision maker, Utility optimizer, Utility-based route planner, Strategic layer planner, Utility-based treatment optimizer, Utility-Based Decision Maker

Variant of Decision Engine abstract

When to choose. Choose when decisions involve complex trade-offs among multiple objectives with no clear priority ordering, probabilistic outcomes, or continuous optimization where the degree of success matters more than binary goal achievement; use scalarized weights when a single decision-maker has clear preferences.

Direct neighbourhood (hover for relationship types)

Relationships

is invoked by dependency

reads dependency

Knowledge Graph Store abstract Ch5.13

writes dependency

Execution Plan abstract Ch5.13

is routed to by dynamic

Hybrid Decision Arbiter Ch5.13

receives data from dynamic

Contextual Weight Adapter Ch5.10

sends data to dynamic

Decision Fusion Aggregator abstract Ch5.13

is constrained by control

Rule Constraint Filter Ch5.13

alternative to variability

Design guidance

SHOULD recalculate expected utilities as sensor data and forecasts update probabilities rather than adding scenario-specific rules.
SHOULD use approximation (sampling, hierarchical decomposition, dominated-action pruning, caching) when exhaustive utility calculation exceeds real-time latency budgets.
SHOULD treat expected-utility rankings as approximate and confirm decisions are robust across plausible probability and utility ranges.
SHOULD NOT conflate utility with monetary or objective value; model diminishing marginal utility explicitly.
SHOULD recompute expected utilities when inputs (e.g., traffic estimates) change, triggering replanning.

Quantitative guidance

As stated by the sources; verify before use.

AV lane change: EU(change lanes) = 0.95 x 0.865 = 0.82 vs EU(stay) = 0.65 with weights 0.5 safety/0.3 speed/0.2 comfort; decision flips if lane-change safety drops to 0.5 or success probability to 0.7 (Ch5.10).
Portfolio example with u(w)=sqrt(w): EU(aggressive)=335.6, EU(balanced)=331.9, EU(conservative)=322.8 (Ch5.10).
Recommendation 0.7 relevance/0.3 diversity utility vs pure relevance: +8% six-month retention, +13% category exploration, -22% 'nothing to watch' reports (Ch5.10).
Utility-based supplier selection: 32% lower annual disruption costs despite 18% higher unit cost vs cost-only; 25% lower cost than premium supplier at equivalent performance (Ch5.10).
Utility-based grid dispatch: 68% fewer unplanned outages vs pure cost minimization; 24% lower operating cost vs continuous spinning reserves (Ch5.10).

Classification

Patterns: Expected utility EU(a) = sum_i P(s_i|a) * u(s_i)Maximum expected utility (MEU) principleMulti-attribute weighted linear utility (scalarization)Sampling-based expected-utility estimationHierarchical action-space decompositionEarly pruning of dominated actionsPortfolio (slate) construction maximizing total expected utility with position-dependent acceptance probabilitiesMean-variance utility u(R) = R - k * sigma^2Quality-adjusted life year (QALY) utilityExpected utility maximizationWeighted multi-objective utility
Quality attributes: Flexibility (ISO/IEC 25010)Functional suitability: correctness and validity (ISO/IEC 25010 | NIST AI RMF: valid)Explainability (NIST AI RMF: explainable and interpretable)Performance efficiency (ISO/IEC 25010)Transparency and accountability (NIST AI RMF: accountable and transparent)Maintainability (ISO/IEC 25010)
Risks mitigated: Unmaintainable rule sets encoding every trade-offBinary goal inflexibilityIgnoring outcome uncertaintyFilter bubbles from pure relevance optimization

Sources

Ch5.10: T. Nguyen, "Utility-Based Decision Making Fundamentals," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 5.10. ISBN: 9798244538229.
Ch5.12: T. Nguyen, "Learning-Based Decision Making Fundamentals," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 5.12. ISBN: 9798244538229.
Ch5.13: T. Nguyen, "Hybrid Decision Systems Fundamentals," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 5.13. ISBN: 9798244538229.