Browse every glossary term as a direct answer page, starting with the exact question readers are likely to ask.
Ensuring AI systems behave in accordance with human values and intentions.
A structured examination of information to identify patterns, meaning, or decisions.
A mechanism that allows models to focus on relevant parts of input when generating output.
Models that generate text one token at a time, using previous tokens to predict the next.
A prompting technique that encourages step-by-step reasoning before providing an answer.
The quality of being easy to understand, precise, and unambiguous in a prompt.
Explicit limits or requirements that narrow how the model should respond.
Information provided to the AI to guide its understanding and response generation.
The maximum amount of text a model can process in a single interaction.
Abbreviation for Chain-of-Thought prompting technique.
Using time, tokens, and computing resources in a way that maximizes useful output.
Moral principles and considerations in AI development and deployment.
The domain knowledge or specialist perspective a prompt asks the model to adopt.
The ability to understand how and why AI systems make decisions.
AI models trained on vast amounts of text data to understand and generate language.
A prompting strategy that starts simple and gradually increases complexity.
The use of valid reasoning patterns to move from premises to conclusions.
Configurable settings that control AI model behavior (temperature, max_tokens, etc.).
The identity, voice, or perspective the model is asked to take on in a prompt.
The input text that instructs an AI model on what to do or generate.
The practice of designing and optimizing prompts to achieve desired AI outputs.
Security vulnerability where malicious input overrides intended prompt instructions.
The degree of variability in a model’s output when multiple valid responses are possible.
A framework combining reasoning and action for complex problem-solving.
The logical process of thinking through problems step by step.
The targeted improvement of a prompt or output based on feedback, testing, or observed failures.
Defining the AI's persona or expertise area in the prompt.
Ensuring AI systems operate without causing harm.
The ability of an AI workflow to handle increasing volume, complexity, or demand without breaking down.
The practice of protecting AI systems and workflows from misuse, manipulation, or unauthorized access.
AI evaluating and improving its own outputs.
Being precise and detailed in prompt instructions.
The logical organization of information within a prompt or response.
Combining multiple inputs, findings, or viewpoints into a coherent unified result.
Instructions that define the AI's behavior and capabilities.
A parameter controlling randomness in AI responses (0 = focused, 1 = creative).
A reusable prompt pattern with placeholders or fixed sections for consistent execution.
The systematic evaluation of prompts and outputs against expected behavior or quality criteria.
The basic unit of text that AI models process (words, parts of words, or characters).
The process of teaching a model by exposing it to data so it can learn patterns and behaviors.
A neural network architecture that uses attention mechanisms to process relationships across sequences of data.
The degree to which an AI system’s behavior, limitations, and decision process are visible and understandable.
Exploring multiple reasoning paths to find the best solution.
Jump into the rest of the prompt engineering guide from the glossary.
Fundamental prompting principles and mental models.
Essential prompting methods and first-line tactics.
More sophisticated reasoning and control patterns.
Structured approaches to building reliable prompts.
Industry-focused prompting approaches and examples.
Production-oriented prompt system design patterns.
Responsible AI use, safeguards, and review practices.
Return to the full prompt engineering guide overview.