Skip to content
AIGP

RFC-034: Domains of Concern for Mediated Intelligence — 4. Core Definitions

AIGP Specification › RFC-034: Domains of Concern for Mediated Intelligence › 4. Core Definitions

← 3. Problem Statement · Section index · 5. Conceptual Model →

4. Core Definitions

4.1 Mediated Intelligence

Mediated intelligence refers to an AI-enabled system that participates in interpretation, recommendation, decision support, action selection, artifact generation, or autonomous execution between a user, context, evidence base, and outcome.

Mediated intelligence may be performed by:

  • a model,
  • an agent,
  • an autonomous system,
  • a workflow,
  • a human-machine team,
  • a governed AI service,
  • a chain of systems.

4.2 Concern

A concern is the specific evaluative question being calculated.

Examples:

  • Was an autonomous action unauthorized?
  • Was a material disclosure misleading?
  • Was a physiological reassurance unsafe?
  • Was a recommendation unsupported by evidence?
  • Was uncertainty hidden from the user?
  • Was a professional judgment made without sufficient competence?
  • Was a cognitive harm trajectory allowed to continue?

A concern must be bounded before it can be calculated.

4.3 Domain of Concern

A Domain of Concern is a bounded evaluative space in which a specific class of mediated intelligence concerns can be empirically observed, measured, and calculated.

A Domain of Concern defines:

  • what concern is in scope,
  • what artifact classes are admissible,
  • what evidence is required,
  • what constraints apply,
  • what uncertainty must be represented,
  • what observer modes are allowed,
  • what variables may be used,
  • what thresholds apply,
  • what governance feedback may follow.

A Domain of Concern is not an industry. It is a constraint-equivalence class.

4.4 Mediation Domain

A Mediation Domain is a reusable class of mediated intelligence usage constraints.

Examples:

  • autonomous systems,
  • regulated advice under uncertainty,
  • evidence-backed material claim generation,
  • professional decision support,
  • risk escalation and triage,
  • longitudinal cognitive interaction,
  • human vulnerability influence,
  • safety-critical action recommendation,
  • high-consequence artifact generation.

A Mediation Domain may appear in many industries.

4.5 Industry Projection

An Industry Projection describes how a Mediation Domain appears inside a specific industry, sector, mission, or institutional setting.

Example:

Mediation Domain:
  Autonomous Systems


Industry Projection:
  Military


Local Use:
  Semi-autonomous reconnaissance drone

The same Mediation Domain may also project into healthcare, cybersecurity, finance, transportation, and industrial operations.

4.6 Local Profile

A Local Profile defines the organization-specific, jurisdiction-specific, artifact-specific, or mission-specific version of a Domain of Concern.

A Local Profile may define:

  • local thresholds,
  • local policy constraints,
  • approval requirements,
  • admissible evidence sources,
  • human review requirements,
  • jurisdictional rules,
  • operational boundaries,
  • escalation paths.

← 3. Problem Statement · Section index · 5. Conceptual Model →