Prompt Engineering refers to methods for communicating effectively with large language models (LLMs) to achieve desired outcomes without updating their underlying weights. This approach, also known as In-Context Prompting, is an empirical science that requires experimentation and heuristics due to its varying effects across different models. The goal of prompt engineering is alignment and model steerability. It focuses on autoregressive language models and aims to control the generation of text. --- Why it matters: Understanding prompt engineering matters to researchers in AI because it can significantly impact the performance and reliability of LLMs, which are increasingly used in applications such as chatbots, language translation, and content generation. Source: https://lilianweng.github.io/posts/2023-03-15-prompt-engineering/