>heyMHK

digital garden

Domains

Notes

select a domain

Home

❯

AI Mechanisms

❯

atoms

Folder: 05-ai-mechanisms/atoms

260 items under this folder.

Jan 06, 2026
Attention as Selective Focus
Jan 03, 2026
Adherence vs Faithfulness vs Groundedness
Jan 03, 2026
Agentic AI
Jan 03, 2026
Agentic RAG
Jan 03, 2026
AI Governance Research Share
Jan 03, 2026
AI Incident Doubling Rate
Jan 03, 2026
AI Redistributes Epistemic Authority
Jan 03, 2026
AI Risk
Jan 03, 2026
AI Risk Measurement Challenges
Jan 03, 2026
AI Risks Differ from Traditional Software Risks
Jan 03, 2026
AI System (Regulatory Definition)
Jan 03, 2026
AI Outperforms Human Prompt Engineering
Jan 03, 2026
Alert Fatigue in Security Operations
Jan 03, 2026
Alert Fatigue in AI-Assisted Systems
Jan 03, 2026
Algorithmic Monoculture
Jan 03, 2026
Asymmetric Miscalibration Failures
Jan 03, 2026
Automation Bias in Regulatory Text
Jan 03, 2026
Behavior and Mechanism Are Interrelated
Jan 03, 2026
When Does an AI State Count as a Belief?
Jan 03, 2026
The Binary Automation Fallacy
Jan 03, 2026
The Brussels Effect in AI Regulation
Jan 03, 2026
The Calibration Detection Gap
Jan 03, 2026
Governance Capacity vs. Governance Effectiveness
Jan 03, 2026
China's Hybrid AI Regulatory Approach
Jan 03, 2026
Coalition Resilience in AI Governance
Jan 03, 2026
Collaboration Channel
Jan 03, 2026
Three Strategies for Agent Collaboration
Jan 03, 2026
Three Types of Agent Collaboration
Jan 03, 2026
Three Communication Structures in Multi-Agent Systems
Jan 03, 2026
Competition Between Agents Improves Output Quality
Jan 03, 2026
Computational Threshold Regulation
Jan 03, 2026
Concise Reasoning Can Match Verbose Reasoning
Jan 03, 2026
Confabulation
Jan 03, 2026
Confabulation vs. Hallucination
Jan 03, 2026
Confidence Calibration
Jan 03, 2026
Confidence Is Not Awareness
Jan 03, 2026
Context Regeneration Filters Irrelevance
Jan 03, 2026
Context Utilization
Jan 03, 2026
Context Window Limitations
Jan 03, 2026
Contextual Disambiguation in Classification
Jan 03, 2026
Coopetition
Jan 03, 2026
Corpus Poisoning Vulnerability
Jan 03, 2026
Deception Requires Intention
Jan 03, 2026
Democratic Integrity as Objective vs. Constraint
Jan 03, 2026
The Demos-to-Deployment Ethics Gap
Jan 03, 2026
Development-Governance Gap Pattern
Jan 03, 2026
Emotional Cues Enhance LLM Performance
Jan 03, 2026
Emotional Cues Enhance LLM Performance
Jan 03, 2026
Ethics Principle Proliferation
Jan 03, 2026
Evaluation Lags Architecture
Jan 03, 2026
Evaluation Metric Limitations
Jan 03, 2026
How Do You Balance Expertise and Democratic Participation in AI Oversight?
Jan 03, 2026
The Expertocracy Problem
Jan 03, 2026
No Single "True" Decomposition Exists
Jan 03, 2026
Exposure Bias
Jan 03, 2026
Factuality Hallucination
Jan 03, 2026
Faithfulness vs Correctness in RAG
Jan 03, 2026
Faithfulness Hallucination
Jan 03, 2026
Few-Shot Chain-of-Thought Consistently Outperforms
Jan 03, 2026
Five Levels of AI Autonomy
Jan 03, 2026
The Fundamental Rights Direct Effect Problem
Jan 03, 2026
GAI Risks Originate Primarily from Human Behavior
Jan 03, 2026
Governance-Development Matching Principle
Jan 03, 2026
GPAI: General-Purpose AI as Regulatory Category
Jan 03, 2026
Quantified Gains from Graduated Human-AI Collaboration
Jan 03, 2026
Core Elements of HAIC Systems
Jan 03, 2026
Three Modes of Human-AI Collaboration
Jan 03, 2026
HITL-Autonomy Inverse Relationship
Jan 03, 2026
The Shift from HMI to HAIC
Jan 03, 2026
The Horizontal-Vertical Regulation Tension
Jan 03, 2026
Horizontal vs Vertical Scaling in AI Systems
Jan 03, 2026
Human Agency and Oversight
Jan 03, 2026
Imitative Falsehood
Jan 03, 2026
In-Context Learning (ICL)
Jan 03, 2026
The Income-Governance Paradox
Jan 03, 2026
Internal vs External Evaluation
Jan 03, 2026
The Knowledge Boundary Problem
Jan 03, 2026
Learning Activities Taxonomy
Jan 03, 2026
Learning Arenas (Governance)
Jan 03, 2026
Learning to Defer (L2D)
Jan 03, 2026
LLM Agreement Predicts Human Agreement
Jan 03, 2026
LLM Annotation Reliability Gap
Jan 03, 2026
The LLM-as-Evaluator Paradigm
Jan 03, 2026
LLMs Excel at Decomposition but Struggle with Computation
Jan 03, 2026
Lying Requires Beliefs and Assertion
Jan 03, 2026
Mechanistic Explanation
Jan 03, 2026
Meta-Learning (Governance)
Jan 03, 2026
Mitigations Must Match Risk Scope
Jan 03, 2026
Model Collapse
Jan 03, 2026
The Overconfident-Wrong Critical Case
Jan 03, 2026
Parametric vs. Non-Parametric Memory
Jan 03, 2026
Prohibited AI Practices (EU Classification)
Jan 03, 2026
Prohibition vs. Permission Regulatory Models
Jan 03, 2026
Prompt Component Taxonomy
Jan 03, 2026
Prompt Engineering
Jan 03, 2026
Prompt Sensitivity Problem
Jan 03, 2026
Prompting vs. Fine-Tuning
Jan 03, 2026
Prompting vs. Prompt Engineering
Jan 03, 2026
Prompts Modify Behavior, Not Knowledge
Jan 03, 2026
Provider vs. Deployer Distinction
Jan 03, 2026
RAG Core Equation
Jan 03, 2026
Retrieval-Augmented Generation (RAG)
Jan 03, 2026
RAG Fusion Strategies: Early vs. Late
Jan 03, 2026
How RAG Reduces Hallucination
Jan 03, 2026
Why Does RAG Still Hallucinate?
Jan 03, 2026
The Four Stages of RAG
Jan 03, 2026
Seven Recurrent RAG Failure Points
Jan 03, 2026
RAG-Specific Attack Vectors
Jan 03, 2026
Reasoning Structure Shapes Capability
Jan 03, 2026
Regulatory Learning
Jan 03, 2026
Reification Gap in LLM Modeling
Jan 03, 2026
Representation Learning for Taxonomy Maintenance
Jan 03, 2026
Residual Risk
Jan 03, 2026
The Retrieval-Generation Cascade Problem
Jan 03, 2026
The Retrieval Noise Paradox
Jan 03, 2026
Retrieval Substitutes for Scale
Jan 03, 2026
Self-Consistency Through Diverse Sampling
Jan 03, 2026
Self-Consistency Shows Limited Effectiveness
Jan 03, 2026
Serious Incident (AI Definition)
Jan 03, 2026
Short-Form Questions Are Answered More Accurately
Jan 03, 2026
Six Factors of Exemplar Effectiveness
Jan 03, 2026
Small Fine-Tuned Models Beat LLM Judges
Jan 03, 2026
Softmax Bottleneck
Jan 03, 2026
The Standards Development Representation Gap
Jan 03, 2026
Static vs Dynamic Orchestration
Jan 03, 2026
Sycophancy Behavior
Jan 03, 2026
What Would It Take to Make Regulatory Learning Systematic?
Jan 03, 2026
Systemic Risk (AI Definition)
Jan 03, 2026
Systemic vs. Individual AI Harms
Jan 03, 2026
Task-Specificity of Prompting Techniques
Jan 03, 2026
TEVV Throughout the AI Lifecycle
Jan 03, 2026
Theory of Mind in Multi-Agent Systems
Jan 03, 2026
Thought as Evaluable Unit
Jan 03, 2026
Three Categories of AI Bias
Jan 03, 2026
Three Configurations of Human-in-the-Loop
Jan 03, 2026
TRACe Framework Decomposition
Jan 03, 2026
Transparency vs. Explainability vs. Interpretability
Jan 03, 2026
The Trust Calibration Problem
Jan 03, 2026
Trust as a Function of Three Factors
Jan 03, 2026
Trustworthiness Tradeoffs Question
Jan 03, 2026
Trustworthy AI Characteristics
Jan 03, 2026
The Uniform Confidence Problem
Jan 03, 2026
The Upstream Evaluation Gap
Jan 03, 2026
Validity and Reliability as Trustworthiness Foundation
Jan 03, 2026
Vehicle vs. Content Distinction
Jan 03, 2026
Verification Before Response Reduces Hallucination
Jan 03, 2026
The Voluntary Compliance Gap
Jan 03, 2026
When Does Retrieval Hurt More Than It Helps?
Jan 02, 2026
The 12B Parameter Threshold
Jan 02, 2026
Agentic AI
Jan 02, 2026
AI Avoids Technology Criticism
Jan 02, 2026
The AI Diversity Deficit
Jan 02, 2026
The AI Execution Gap
Jan 02, 2026
Algorithmic Hubris
Jan 02, 2026
Asymmetric Persona Effects
Jan 02, 2026
Attention Heads Learn Specialized Tasks
Jan 02, 2026
Constant-Time Attention Paths
Jan 02, 2026
Attention Is Weighting, Not Understanding
Jan 02, 2026
The Automation-Augmentation Paradox
Jan 02, 2026
What Benchmarks Exclude
Jan 02, 2026
The CACE Principle
Jan 02, 2026
Capability Evaluations as Leading Indicators
Jan 02, 2026
Co-Evolution (Human-AI)
Jan 02, 2026
Configuration Debt
Jan 02, 2026
Configuration Sensitivity in Modular AI Systems
Jan 02, 2026
Context Reduction Improves LLM Precision
Jan 02, 2026
The Context Specification Gap
Jan 02, 2026
Correction Cascades
Jan 02, 2026
The Chain-of-Thought Diversity Effect
Jan 02, 2026
Demographic Persona Conditioning Is Necessary for Signal
Jan 02, 2026
The Deploy/Maintain Dichotomy
Jan 02, 2026
Detection Tools Cover Only 45% of Known Pattern Types
Jan 02, 2026
The Direct Likert Regression Problem
Jan 02, 2026
The Empathy Paradox
Jan 02, 2026
Empowerment Parity Across Retrieval Architectures
Jan 02, 2026
The Excessive Automation Danger Zone
Jan 02, 2026
Expert Parity Trajectory
Jan 02, 2026
Few-Shot Learning Plateaus Fast
Jan 02, 2026
Five Dimensions of Information Completeness
Jan 02, 2026
Flaws as Authenticity Signals
Jan 02, 2026
Global Sensemaking vs. Local Retrieval
Jan 02, 2026
Hallucination as Untraceable Accuracy
Jan 02, 2026
Hallucination from Ungrounded Reasoning
Jan 02, 2026
Hidden Feedback Loops
Jan 02, 2026
Idea Exhaustion Dynamics
Jan 02, 2026
Idea Space as Rugged Landscape
Jan 02, 2026
Inner Speech and Cognitive Self-Regulation
Jan 02, 2026
Input Modality Shapes What Detection Systems Can Find
Jan 02, 2026
Knowledge Graphs as Dynamic Memory Substrate
Jan 02, 2026
Knowledge Spreads Like Disease
Jan 02, 2026
LLMs as Approximate Knowledge Bases
Jan 02, 2026
LLM-as-Oracle vs LLM-as-Aligner
Jan 02, 2026
LLM Consensus as Quality Proxy
Jan 02, 2026
LLM Oracles Match 20% Error Rate Human Experts
Jan 02, 2026
LLM Stereotype Defaults
Jan 02, 2026
Different LLMs Encode Different Subjective Biases
Jan 02, 2026
LLM Validation: $0.01-$0.04 per 1,000 Assessments
Jan 02, 2026
LLMs Follow Patterns Better Than Instructions
Jan 02, 2026
LLMs Over-Generate Structured Output
Jan 02, 2026
Low Between-Strategy Overlap
Jan 02, 2026
The ML Code Fraction
Jan 02, 2026
Model Entanglement
Jan 02, 2026
Model Failure Mode Distribution
Jan 02, 2026
Model Strengths Differ by Modality
Jan 02, 2026
Model Valorization
Jan 02, 2026
Modularity Unlocks LLM Performance on Complex Tasks
Jan 02, 2026
Multi-Head Attention
Jan 02, 2026
Multi-Hop Reasoning
Jan 02, 2026
"In real life, we never see clean data"
Jan 02, 2026
Nine-Stage ML Workflow
Jan 02, 2026
Natural Language Prompts Outperform Structured Prompts
Jan 02, 2026
Non-Monotonic Error Propagation
Jan 02, 2026
o1-Preview Achieves Expert-Level Ontology Modeling
Jan 02, 2026
Occupation-Hobby Alignment Tells
Jan 02, 2026
The One-Directional Communication Gap
Jan 02, 2026
You Don't Need to Find the Expert
Jan 02, 2026
Persona Beats Methodology
Jan 02, 2026
Persona-Capability Dissociation
Jan 02, 2026
Persona Drift
Jan 02, 2026
Physical World Brittleness
Jan 02, 2026
The Polish-Authenticity Paradox
Jan 02, 2026
Positional Encoding
Jan 02, 2026
Proactive vs. Reactive Clarification
Jan 02, 2026
Query-Key-Value Framework
Jan 02, 2026
ReAct Benchmark Results
Jan 02, 2026
The Flexibility-Factuality Tradeoff
Jan 02, 2026
Role Constraint Refusal
Jan 02, 2026
RST: Reliable, Safe & Trustworthy
Jan 02, 2026
Safety Guardrails as Domain Modeling Blind Spots
Jan 02, 2026
Self-Attention
Jan 02, 2026
LLM Self-Critique Tracks External Quality
Jan 02, 2026
The Sequential Bottleneck Problem
Jan 02, 2026
Single-Turn Comprehensiveness
Jan 02, 2026
Sparse vs. Dense Reasoning
Jan 02, 2026
Symbol Grounding Problem
Jan 02, 2026
Synonyms Beat Hierarchy in Ontology Prompts
Jan 02, 2026
The 94% Threshold
Jan 02, 2026
Three Fundamental ML Engineering Differences
Jan 02, 2026
The Toy Dataset Training Gap
Jan 02, 2026
Undeclared Consumers
Jan 02, 2026
Vector vs Graph RAG Performance Gap
Jan 02, 2026
Vector RAG Directness Win Rate: 92%
Jan 02, 2026
Visual Self-Checking Improves Performance
Jan 01, 2026
AI Agent (Occupational Context)
Jan 01, 2026
Attention Accelerates Training Without Improving Accuracy
Jan 01, 2026
The Benchmark-to-Production Gap in NLP
Jan 01, 2026
Common Words Matter More Than You'd Expect
Jan 01, 2026
The Consensus Erosion Pattern
Jan 01, 2026
Constraints Beat Personas
Jan 01, 2026
Context Position Matters: The Middle Gets Lost
Jan 01, 2026
Custom Embeddings Outperform Generic Ones
Jan 01, 2026
Custom NER Dramatically Outperforms Open-Source in Domain-Specific Text
Jan 01, 2026
Far-Field Multi-Speaker Audio Is the Hardest ASR Problem
Jan 01, 2026
Fine-Tuning Shifts Attention, Not Knowledge
Jan 01, 2026
H-Score
Jan 01, 2026
Hallucination Is Inherent, Not a Bug Being Fixed
Jan 01, 2026
Highlight Extraction vs. Text Summarization
Jan 01, 2026
Label Distillation
Jan 01, 2026
Physical vs. Nonphysical Governance
Jan 01, 2026
Punctuation Dramatically Improves Sequential NLP Models

hey-mhk.com