Concept Taxonomy¶
This document defines the categorical taxonomy for organizing the 200 concepts in the SEIS 666 learning graph.
Taxonomy Categories¶
| Category Name | TaxonomyID | Description |
|---|---|---|
| Foundation Concepts | FOUND | Core digital transformation and AI foundational concepts |
| LLM Architecture | ARCH | Technical architecture concepts for large language models |
| AI Platforms | PLAT | Specific AI platforms, products, and services |
| Prompt Engineering | PROMPT | Techniques and methods for prompting LLMs |
| Custom Solutions | CUSTOM | Custom GPTs, agents, and no-code AI tools |
| API Integration | API | Technical concepts for API integration |
| Multimodal AI | MULTI | Text-to-image, vision, audio, and other modalities |
| Governance | GOV | AI governance, strategy, and organizational excellence |
| Ethics | ETHICS | Responsible AI, bias, privacy, and compliance |
| Workforce | WORK | Future of work and workforce transformation |
| Business Applications | BIZ | Use cases, case studies, and business applications |
| Advanced Topics | ADV | Converging technologies and advanced implementations |
Category Definitions¶
FOUND - Foundation Concepts¶
Core concepts that establish the fundamentals of digital transformation and artificial intelligence. These are typically foundational prerequisites that many other concepts build upon.
Examples: Digital Transformation, Digitization, Digital Maturity, Artificial Intelligence, Machine Learning
ARCH - LLM Architecture¶
Technical concepts related to the architecture, training, and operation of large language models. Understanding these concepts helps explain how LLMs work.
Examples: Transformer Architecture, Attention Mechanism, Pre-Training, Fine-Tuning, RLHF, Token, Context Window
PLAT - AI Platforms¶
Specific AI platforms, products, and services from various providers. These concepts represent the commercial and open-source tools students will use.
Examples: OpenAI, GPT-4, ChatGPT, Anthropic, Claude, Google Gemini, Perplexity AI, Llama, Mistral
PROMPT - Prompt Engineering¶
Techniques, strategies, and methods for effectively prompting large language models to achieve desired outputs.
Examples: Zero-Shot Prompting, Few-Shot Prompting, Chain-of-Thought, System Prompt, Output Formatting
CUSTOM - Custom Solutions¶
Building custom AI solutions including custom GPTs, AI agents, and leveraging no-code/low-code platforms.
Examples: Custom GPT, GPT Builder, AI Agents, Autonomous Systems, No-Code AI Tools, RAG
API - API Integration¶
Technical concepts for integrating with LLM APIs, including authentication, parameters, and optimization.
Examples: REST API, OpenAI API, API Authentication, Temperature Parameter, Rate Limiting
MULTI - Multimodal AI¶
AI capabilities beyond text, including image generation, vision, audio, and video.
Examples: Text-to-Image, DALL-E, Midjourney, Diffusion Models, Vision Capabilities, Audio AI
GOV - Governance¶
Organizational concepts for AI governance, strategy development, and establishing AI Centers of Excellence.
Examples: GAI Center of Excellence, AI Governance, AI Strategy, Change Management, Scaling AI
ETHICS - Ethics¶
Responsible AI principles, bias mitigation, privacy, security, and regulatory compliance.
Examples: AI Bias, Bias Mitigation, Hallucination, Data Privacy, Responsible AI, Red-Teaming
WORK - Workforce¶
Future of work concepts addressing how AI transforms jobs, skills, and organizational structures.
Examples: AI-Augmented Workforce, Skill Transformation, Role Evolution, Human-AI Collaboration
BIZ - Business Applications¶
Practical business applications including use case identification, prioritization, and industry examples.
Examples: AI Use Case, Use Case Identification, ROI Estimation, Healthcare AI, Case Study Analysis
ADV - Advanced Topics¶
Advanced and emerging topics including converging technologies and strategic transformation.
Examples: Converging Technologies, Edge AI, AI Infrastructure, AI Transformation, Capstone Project
Taxonomy Distribution Target¶
To ensure balanced coverage, each category should ideally contain:
- Minimum: 10 concepts (~5%)
- Maximum: 35 concepts (~17.5%)
- Target average: ~17 concepts per category
Categories with significantly more or fewer concepts should be reviewed for potential splitting or merging.