DataTerminology.comNEW
The practice of designing and refining inputs to guide generative AI models toward desired outputs.
DESCRIPTION
Prompt engineering crafts instructions, context, and examples to steer large language and generative models effectively. Techniques include role prompting, chain-of-thought, few-shot examples, and tool or retrieval augmentation. It balances specificity with flexibility, often coupled with evaluation harnesses and guardrails. Security concerns include prompt injection and data leakage. As models evolve, prompt patterns and libraries help standardize reliable interactions.
ALTERNATIVE DEFINITIONS
Wikipedia
Prompt engineering is the process of crafting inputs that can be interpreted and acted on by a generative AI model.
General definition tying prompts to generative model behavior.
OpenAI
Prompt engineering is the practice of designing inputs for LLMs to reliably achieve desired outputs.
Vendor guidance focused on reliability of LLM outputs.
Google Cloud
Prompt design is the process of creating inputs to help generative AI models produce relevant, high-quality outputs.
Emphasizes relevance and quality in multimodal contexts.
RELATED TERMS
Large Language Model Instruction Tuning Chain-of-Thought Few-Shot Learning Assistant Prompt Prompt Injection Guardrails Evaluation
COMPETING TERMS
Fine-Tuning Retrieval-Augmented Generation
RELATIONS
- Narrower: Chain-of-Thought
- Referenced by: Berkeley Function-Calling Leaderboard
KNOWLEDGE AREAS
TARGET AUDIENCE
AI Engineer MLOps Engineer Data Scientist AI Product Manager Software Developer
EXTERNAL SOURCES
- Wikipedia - Prompt engineering
- OpenAI - LLM prompt engineering best practices
- Microsoft Learn - Prompt patterns and anti-patterns
- Google Cloud - Prompt design for generative models
TOP QUESTIONS
What is Prompt Engineering?
It’s designing inputs—instructions, context, and examples—to guide generative models to produce desired, reliable outputs.
Category: Definition
When should I prefer Prompt Engineering over Fine-tuning?
Start with prompt engineering for flexibility and speed; fine-tune when consistent domain-specific behavior is required.
Category: Opinion
What are common techniques in Prompt Engineering?
Chain-of-thought, few-shot examples, role prompts, and tool or retrieval augmentation are widely used.
Category: Techniques
How do I evaluate prompts?
Use test suites with ground truth, human review, and automated metrics; iterate to balance quality and cost.
Category: Testing
What risks exist in Prompt Engineering?
Prompt injection, data leakage, and brittle prompts are risks. Guardrails and content filters help mitigate them.
Category: Risks
Does Prompt Engineering generalize across models?
Patterns often transfer, but specifics vary by model. Validate prompts across targets when portability matters.
Category: Misunderstanding
How does RAG relate to Prompt Engineering?
RAG augments prompts with retrieved context to ground outputs in authoritative sources for accuracy.
Category: Comparison
What is a system prompt?
A persistent instruction that sets behavior and style for the model across a conversation or session.
Category: Definition
What is trending in Prompt Engineering?
Programmatic prompting, prompt templates, and evaluation harnesses integrated into CI pipelines are rising.
Category: Topic