Framework Reference

AI Ethics vs.
AI Governance

Ethics is about reasoning and values. Governance is about machinery and rules. Neither works without the other. Explore how they're distinct — and why they must converge.

The distinction
01
AI Ethics
The Reasoning Layer
Values, principles, moral judgment, and consequences. Defines the boundaries of acceptable action.
Goal: Doing the right thing.
↔
Responsible AI
The Essential Synergy
Where values become enforceable. Neither layer is sufficient alone — both are required for genuine accountability.
Goal: Making it stick.
02
AI Governance
The Operational Layer
Rules, processes, roles, audits, and accountability structures. Builds the practical machinery to enforce values.
Goal: Making it happen.
Core Concern: What is right?
What is fair? What counts as harm? Should we use AI for this at all? What tradeoffs between competing goods — accuracy vs. privacy, speed vs. safety — are acceptable?
What ethics produces
Principles, ethical analyses, human-impact frameworks, and the reasoning behind organizational values. Ethics is where you discover what you're committed to before any rules are written.
What each covers
AI Ethics
Values Foundation
“What is right?”
FairnessHarmAutonomyTradeoffsDignityPurpose
Shared
Responsible AI
Click to explore
AI Governance
Operational System
“How do we manage it?”
Access ControlAuditAccountabilityPolicyIncident Response
Core questions
  • —What is fair — and to whom?
  • —What counts as harm in this context?
  • —Should we use AI for this purpose at all?
  • —How do we weigh accuracy vs. privacy?
  • —Are we respecting individual autonomy and dignity?
Outputs
  • —Ethical principles and value statements
  • —Human-impact frameworks
  • —Stakeholder analysis and harm mapping
  • —Tradeoff analyses for specific use cases
In practice — AI integration

Ethical Context Engineering

Responsible AI integration requires managing how information flows into and out of AI systems. This four-stage pipeline shows how ethical constraints become technical controls. Click any stage to examine its role.

Stage 01
⚙️
System Prompts
Foundation & behavioral guardrails
Stage 02
🔍
Controlled Retrieval
Data intake & RAG constraints
Stage 03
🗜️
Context Compaction
Focus & drift prevention
Stage 04
🛡️
Output Guardrails
Validation & auditability
Methodology
Establish behavioral guidelines and value-aligned guardrails at the structural level using system prompts. Ethical principles — fairness, harm avoidance, transparency — are embedded at the model's foundation before any user interaction occurs.
Impact
Embeds core principles directly into the model's reasoning layer, ensuring it resists adversarial inputs and adheres to safety constraints across all downstream interactions. This is where ethics becomes architecture.
In practice
💼AI Hiring ScreeningApplicant filtering at scale
🏥Hospital Patient RiskPredicting patient deterioration
🎓Faculty Using LLMsAI in course design and assessment
Ethical Analysis

Before deploying: what must the organization ask?

  • —Is it fair to use an algorithm to filter human potential?
  • —Does this tool systematically exclude protected groups?
  • —Are we avoiding disability discrimination in training data?
  • —Do candidates deserve an explanation when AI affects their future?
Governance Action

What rules and structures must be in place?

  • —Define which data fields can and cannot feed the model
  • —Require HR sign-off before any deployment decision
  • —Mandate quarterly bias audits with documented results
  • —Build an appeal process into the hiring workflow
The responsible AI ecosystem

Ethics and governance are two of three interlocking disciplines that together define responsible AI. Click any node to explore its role.

The Goal
Responsible AI
AI Ethics
The Values
Fairness, harm, autonomy, and the moral questions that must be asked before any system is built or deployed.
AI Governance
The Machinery
Policies, review boards, audit workflows, and accountability structures that make ethical commitments enforceable.
Technical Robustness
The Engineering
Model security, red-teaming, agentic safeguards, and the technical controls that make systems resistant to failure and misuse.

This framework is drawn from the Applied AI Ethics curriculum. Ethics defines what we're committed to. Governance makes that commitment traceable. Technical robustness makes it real. The Discovery Assessment maps how individuals reason across these tensions — in real tradeoffs, not abstractions.

The Framework — Applied AI Ethics Project | Applied AI Ethics Project