AI Fluency & Technical Concepts

44 vocabulary flashcards covering key terms from the AI Fluency framework, human-AI interaction modes, technical concepts, and prompt-engineering techniques.

AI Fluency

The capacity to work with AI systems effectively, efficiently, ethically, and safely, combining practical skills, knowledge, insights, and values.

The 4Ds

The four core AI-fluency competencies: Delegation, Description, Discernment, and Diligence.

Delegation

Choosing which tasks humans do, which tasks AI does, and how to divide work based on goals and capabilities.

Problem Awareness

Clearly understanding goals and task requirements before involving AI.

Platform Awareness

Knowing the capabilities and limitations of different AI systems.

Task Delegation

Thoughtfully distributing work between humans and AI to leverage each party’s strengths.

Description

Communicating with AI by precisely defining outputs, processes, and desired behaviors.

Product Description

Stating desired output, format, audience, and style for the AI.

Process Description

Giving step-by-step guidance on how the AI should approach a request.

Performance Description

Specifying how the AI should behave (e.g., concise, detailed, supportive, challenging).

Discernment

Critically evaluating AI outputs, processes, and behaviors for quality and improvement.

Product Discernment

Assessing the accuracy, relevance, coherence, and appropriateness of AI outputs.

Process Discernment

Checking the reasoning steps the AI used for errors or lapses.

Performance Discernment

Judging whether the AI’s communication style meets your needs.

Diligence

Using AI responsibly and ethically with transparency and accountability.

Creation Diligence

Choosing AI systems thoughtfully and interacting with them responsibly.

Transparency Diligence

Being open about AI’s role with everyone who needs to know.

Deployment Diligence

Verifying and vouching for any AI-assisted outputs you use or share.

Automation

AI performs specific tasks exactly as instructed by a human.

Augmentation

Humans and AI iterate together as thinking partners to complete tasks.

Agency

Humans configure AI to act independently on their behalf, even interacting with others.

Generative AI

AI systems that create new content (text, images, code, etc.) rather than merely analyzing data.

Large language models (LLMs)

Generative AI trained on massive text corpora to understand and generate human language.

Claude

Anthropic’s family of large language models.

Parameters

Numerical values in a model that determine how it processes information; modern LLMs have billions.

Neural networks

Layered collections of interconnected nodes that learn patterns from data through training.

Transformer architecture

2017 breakthrough design enabling models to process text in parallel and attend across long passages.

Scaling laws

Empirical patterns showing model performance improves predictably with more data, compute, and size, often unlocking new abilities.

Pre-training

Initial training phase where models learn language patterns from vast text data.

Fine-tuning

Additional training that teaches models to follow instructions and avoid harmful content.

Context window

Maximum amount of information a model can consider at once, including conversation history and documents.

Hallucination

Error where AI confidently provides plausible-sounding but incorrect information.

Knowledge cutoff date

The latest point in time covered by a model’s training data; it lacks built-in knowledge after this date.

Reasoning or thinking models

AI models designed to work step-by-step through complex problems, improving logical reasoning tasks.

Temperature

Setting that controls randomness of AI responses—higher is more creative, lower more predictable.

Retrieval-augmented generation (RAG)

Technique that links models to external knowledge sources to boost accuracy and reduce hallucinations.

Bias

Systematic patterns in outputs that unfairly favor or disadvantage certain groups, often reflecting training data.

Prompt

Any input given to an AI model, including instructions and shared documents.

Prompt engineering

Designing effective prompts to obtain desired outputs from AI systems.

Chain-of-thought prompting

Asking AI to reason through a problem step by step.

Few-shot learning (n-shot prompting)

Teaching AI by providing N example input-output pairs to illustrate the desired pattern.

Role or persona definition

Directing the AI to adopt a specific character, expertise level, or communication style.

Output constraints / output formatting

Specifying format, length, structure, or other characteristics required in the AI’s response.

Think-first approach

Prompting the AI to work through its reasoning before giving a final answer, leading to deeper analysis.