Skip to main content
Atlas Manual

AI Alignment & Incentives Catalog

85 terms across 9 themes — every definition in the Atlas sourced from this manual, with executive examples and counter-moves.

85 of 85 terms
AI Incentives · Governance
Acceptable Use Policy (AI)

Written rules about how AI may be used internally.

AI Incentives · Adoption
Adoption Curve (AI)

Innovators → early adopters → majority → laggards, AI-specific.

AI Incentives · Workflow
Agent

AI system that takes actions to achieve goals, often across tools.

AI Incentives · Risk
Agentic Liability

When the agent acts, who's responsible?

AI Incentives · Strategy
AI as Coach vs. AI as Replacement

Augmentation strategy vs. substitution strategy.

AI Incentives · Governance
AI Audit Trail

Logging of AI inputs, outputs, and decisions.

AI Incentives · Governance
AI Bill of Materials

Inventory of models, data, tools, and dependencies in an AI system.

AI Incentives · Governance
AI Bill of Materials Auditability

Tracing AI components for risk and compliance.

AI Incentives · Governance
AI Council / Committee

Cross-functional governance body for AI decisions.

AI Incentives · Governance
AI Governance Vacuum

Agents deployed before anyone owns the consequences.

AI Incentives · Strategy
AI Maturity Model

Stages of organizational AI capability.

AI Incentives · Economics
AI Productivity Paradox

AI investment outpacing measurable productivity gains.

AI Incentives · Risk
AI Risk Tiering

Categorizing AI use cases by risk level.

AI Incentives · Strategy
AI Strategy as Capability Strategy

AI strategy = decisions about which capabilities to build and where.

AI Incentives · Talent
AI Talent Premium

Scarce AI talent commands market-distorting compensation.

AI Incentives · Workflow
AI-First Workflow Design

Designing processes from scratch around AI capability.

AI Incentives · Talent
AI-Native vs. AI-Augmented Teams

Teams built around AI from day one vs. teams adding it to existing workflows.

AI Incentives · Adoption
Algorithmic Aversion

Discounting algorithmic advice even when superior.

AI Incentives · Trust
Anthropomorphism

Treating AI as more humanlike than it is.

AI Incentives · Strategy
Augmentation vs. Automation

Augment when judgment matters. Automate when scale matters.

AI Incentives · Trust
Automation Complacency

Reduced vigilance with automated systems.

AI Incentives · Risk
Bias in AI Systems

Systematic skew in model behavior across groups.

AI Incentives · Strategy
Build vs. Buy vs. Partner (AI)

Strategic choice on AI capability sourcing.

AI Incentives · Strategy
Capability Overhang

AI capability outpacing organizational ability to use it.

AI Incentives · Adoption
Cargo-Cult Adoption

AI bolted onto existing workflows to look forward-leaning.

AI Incentives · Economics
Compute Economics

GPU access and pricing shape what's feasible.

AI Incentives · Workflow
Concept Drift

The relationship between inputs and outputs changes.

AI Incentives · Trust
Confidence Intervals (AI)

Quantifying uncertainty in model outputs.

AI Incentives · Alignment
Constitutional AI

Models trained to follow a written set of principles.

AI Incentives · Governance
Content Provenance (C2PA)

Cryptographic tracking of content origin.

AI Incentives · Workflow
Context Window

How much input the model can process at once.

AI Incentives · Economics
Cost-Per-Token Strategy

Operating economics shaped by per-token pricing.

AI Incentives · Risk
Counterfactual Testing

Testing model behavior on hypothetical alternate inputs.

AI Incentives · Workflow
Data Drift

Underlying data distribution changes over time.

AI Incentives · Governance
Data Lineage

Tracking where training and inference data came from.

AI Incentives · Risk
Differential Privacy

Adding noise to data to protect individual privacy.

AI Incentives · Workflow
Embedding

Vector representation of content for similarity and search.

AI Incentives · Governance
EU AI Act

Comprehensive AI regulation in the EU.

AI Incentives · Workflow
Eval Drift

Evaluation suites that no longer reflect real-world conditions.

AI Incentives · Workflow
Evaluation Frameworks (Evals)

Systematic testing of model quality, safety, and capability.

AI Incentives · Governance
Explainability vs. Interpretability

Why did it produce this? vs. How does it work?

AI Incentives · Risk
Fairness Metrics

Quantitative measures of model behavior across groups.

AI Incentives · Risk
Federated Learning

Training models across devices without centralizing data.

AI Incentives · Strategy
Fine-Tuning

Specialized training on domain data.

AI Incentives · Economics
Foundation Model Concentration

A handful of providers shape the entire AI economy.

AI Incentives · Alignment
Goodhart's Law (AI form)

Optimizing a proxy of the goal degrades the actual goal.

AI Incentives · Trust
Hallucination

Confident outputs that are factually wrong.

AI Incentives · Governance
Human-in-the-Loop

Human review at critical AI decision points.

AI Incentives · Governance
Human-on-the-Loop

Human oversight without per-decision review.

AI Incentives · Workflow
In-Context Learning

Models learning from examples in the prompt.

AI Incentives · Economics
Inference vs. Training Cost

Training is a one-time cost. Inference is forever.

AI Incentives · Alignment
Inner vs. Outer Alignment

Outer: the spec matches our intent. Inner: the model actually pursues the spec.

AI Incentives · Risk
Jailbreak

Bypassing model safety constraints.

AI Incentives · Adoption
Job Redesign

Redesigning roles around AI capability.

AI Incentives · Workflow
Latency Budget

How much delay the user experience tolerates.

AI Incentives · Alignment
Mesa-Optimization

The trained model develops its own internal optimizer.

AI Incentives · Governance
Model Cards

Documentation of model purpose, performance, limitations, and risks.

AI Incentives · Workflow
Model Distillation

Train a smaller model to imitate a larger one.

AI Incentives · Workflow
Model Drift

Performance degradation as real-world data shifts.

AI Incentives · Workflow
Multi-Agent Systems

Multiple specialized agents coordinating on tasks.

AI Incentives · Governance
NIST AI RMF

U.S. voluntary AI risk management framework.

AI Incentives · Alignment
Objective Function

What the model is actually optimizing.

AI Incentives · Workflow
On-Device vs. Cloud

Where the model runs shapes privacy, latency, cost, and capability.

AI Incentives · Strategy
Open vs. Closed Models

Open-weight vs. API-only models.

AI Incentives · Adoption
Pilot Purgatory

AI pilots that succeed and never scale.

AI Incentives · Risk
Productivity Mirage

Individual speed gains hide collective quality decline.

AI Incentives · Economics
Prompt Caching

Cache common prompt prefixes to reduce cost and latency.

AI Incentives · Risk
Prompt Injection

Malicious instructions hidden in user input or retrieved content.

AI Incentives · Risk
Red-Teaming AI

Adversarial testing of AI systems.

AI Incentives · Adoption
Resistance Mapping

Pre-deployment analysis of who loses what.

AI Incentives · Talent
Reskilling vs. Replacing

Investing in workforce transition vs. workforce change.

AI Incentives · Workflow
Retrieval-Augmented Generation (RAG)

Generate based on retrieved documents, not just trained weights.

AI Incentives · Alignment
Reward Hacking

Maximizing the reward signal in unintended ways.

AI Incentives · Alignment
RLHF

Reinforcement learning from human feedback.

AI Incentives · Adoption
Sandbagging Adoption

Teams under-reporting AI capability to protect comp or status.

AI Incentives · Adoption
Shadow AI

Employees using unauthorized AI tools to get work done.

AI Incentives · Risk
Skill Atrophy

Foundational skills erode through AI offloading.

AI Incentives · Alignment
Specification Gaming

The model achieves the goal as stated, not as intended.

AI Incentives · Adoption
Status Threat (AI)

AI doesn't just replace tasks — it threatens identities.

AI Incentives · Workflow
Synthetic Data

Generated training data that mimics real data.

AI Incentives · Workflow
Tool Use

Models calling external functions and APIs.

AI Incentives · Economics
Total Cost of Ownership (AI)

Real cost includes data, ops, monitoring, governance, training.

AI Incentives · Trust
Trust Calibration

Matching trust in a system to its actual reliability.

AI Incentives · Workflow
Vector Database

Database optimized for high-dimensional similarity search.

AI Incentives · Strategy
Vendor Lock-In Risk

Concentration risk on a single AI provider.