Prompt
The setup + the question you send the model — everything inside the input box. A well-structured prompt is the difference between a hallucination and a useful, compliant summary.
The input provided to an LLM to elicit a response. It consists of the system instructions, user input, and any additional context or examples provided to guide the model's output.
The input provided to an LLM to elicit a response. It consists of the system instructions, user input, and any additional context or examples provided to guide the model's output.
evolution
- 1960 · historyELIZA and Scripted Inputs
Early chatbots like ELIZA introduced the concept of user-provided text strings to trigger specific, rule-based conversational responses.
- 2017-06 · historyTransformer Architecture
The 'Attention Is All You Need' paper established the mechanism for models to process entire input sequences simultaneously, setting the stage for modern prompt engineering.
- 2020-05 · historyGPT-3 Few-Shot Prompting
The release of GPT-3 demonstrated that models could perform tasks by providing examples within the prompt, formalizing 'few-shot' prompting as a primary interaction method.
- 2022-11 · historyChatGPT and Conversational Prompting
The launch of ChatGPT popularized iterative, natural language prompting as a standard interface for interacting with Large Language Models.
- 2023-01 · historyEmergence of Prompt Engineering
Prompt engineering became recognized as a distinct technical discipline focused on optimizing input structure to improve model reasoning and output accuracy.