Practical Applications of Prompt - D685

This set of vocabulary flashcards covers AI fundamentals, prompt engineering frameworks, prompting techniques, limitations, and ethical principles based on the D685 course study guide.

Artificial Intelligence (AI)

The simulation of human intelligence processes by machines, especially computer systems, including learning, reasoning, and self-correction.

Narrow AI (Weak AI)

AI designed to perform a specific task or narrow set of tasks, such as chatbots or facial recognition, that cannot perform tasks outside its programmed scope.

General AI (Strong AI)

A hypothetical future AI system that would possess human-level intelligence and could perform any intellectual task that a human can do; it does not currently exist.

Machine Learning (ML)

A subset of AI where computers learn from data without being explicitly programmed, improving performance through experience.

Supervised Learning

A machine learning technique where a model is trained using labeled data with known outcomes to make predictions on new data.

Unsupervised Learning

A type of machine learning where the model is trained on unlabeled data and must find patterns and relationships independently.

Reinforcement Learning

A machine learning approach where an AI agent learns through trial and error, receiving rewards or penalties to maximize rewards over time.

Neural Networks

Computing systems inspired by biological networks consisting of interconnected layers of artificial neurons to recognize patterns.

Deep Learning

A subset of machine learning using neural networks with multiple layers to analyze data, effective for image recognition and NLP.

Large Language Models (LLMs)

AI systems trained on massive text data to understand and generate human-like language, such as ChatGPT, Claude, and GPT-4.

Natural Language Processing (NLP)

The field of AI focused on enabling computers to understand, interpret, and generate human language through applications like sentiment analysis and translation.

Computer Vision

A field of AI that enables computers to interpret and analyze visual information from images and videos for tasks like object detection.

Generative AI

AI systems capable of creating new content such as text, images, audio, or video based on provided inputs.

Training Data

The dataset used to teach a machine learning model, where quality and size significantly impact performance and bias.

Tokens

The fundamental units of text processed by LLMs, where $1$ token is roughly equal to $4$ characters or $0.75$ words in English.

Token Limit

The maximum number of tokens a model can process in a single interaction, affecting context window and memory constraints.

Temperature

A parameter in LLMs controlling randomness/creativity, where low values ($0.1-0.5$) are predictable and high values ($0.7-1.0$) increase variety.

Prompt Engineering

The practice of designing and refining prompts to achieve desired outputs from AI systems by maximizing their effectiveness.

PIIOA Framework

The five essential components of an effective prompt: Persona, Instructions, Input, Output format, and Additional information.

Persona

The role or identity assumed by the AI, which influences the tone, style, and expertise level of its responses.

Specificity

The degree to which prompts are tailored to address particular needs and context; higher specificity leads to more accurate responses.

Context

The surrounding circumstances or background information that helps the AI understand the situation behind a request.

Constraints

Limitations or rules that restrict AI behavior, such as word limits, format requirements, or tone specifications.

Verbosity

The level of detail or conciseness in a prompt's language.

Zero-Shot Prompting

Providing no examples and relying on the model's pre-existing knowledge for direct or factual queries.

Few-Shot Prompting

Providing $2-5$ examples to guide the AI's format and style for consistent output patterns.

Chain-of-Thought (CoT)

Requesting step-by-step reasoning from the AI to help it 'show its work' during complex problem-solving or logic tasks.

Tree-of-Thought (ToT)

Exploring multiple solution paths before responding, useful for creative tasks and evaluating alternatives.

Cognitive Verifier Pattern

A technique where the AI asks clarifying questions first to deepen understanding before responding to ambiguous requests.

Least-to-Most Prompting

Gradually increasing complexity by building from simple concepts to advanced ones.

Iterative Refinement

The process of starting with a basic prompt and progressively adding specificity and context until the desired result is achieved.

Sentiment Analysis

The process of determining the emotional tone (positive, negative, or neutral) expressed in text.

DALL·E

An AI tool used for text-to-image generation with high creativity.

OCR (Optical Character Recognition)

Technology that converts printed or handwritten text into digital, editable text.

Hallucination

When AI generates false, fabricated, or nonsensical information presented confidently as fact.

Knowledge Cutoff

The date beyond which an AI model has no information, determined by when its training data ends.

Context Window

The maximum amount of text (measured in tokens) an AI model can 'remember' or process at one time.

Sampling Bias

Bias that occurs when training data does not represent the whole population, leading to skewed results.

Measurement Bias

Errors or inconsistencies in data collection methods that systematically affect results, such as low-quality sensors.

Algorithmic Bias

When the design of a model itself perpetuates or amplifies bias, even if the input data is fair.

Selection Bias

Bias caused by non-random data selection that leaves gaps in representation, such as excluding failures.

Confirmation Bias

Training AI only on confirming examples that support existing assumptions rather than challenging them.

AI Ethics

The field concerned with ensuring AI systems are developed and used responsibly, fairly, and in alignment with human values.

Fairness

Ensuring equitable treatment across all groups and actively mitigating discriminatory outcomes in AI systems.

Accountability

Establishing clear responsibility for AI decisions and outcomes, including mechanisms for recourse when mistakes occur.

Transparency

Making AI decision-making processes explainable and understandable so users know how conclusions are reached.

Societal Impact

The ethics principle of considering broader effects of AI, such as employment impacts, inequality, and the digital divide.