**Prompt engineering** is the process of applying prompting to effectively extract information from language models, typically for use in real-world applications. Correct prompt engineering is coupled with a process of controlled testing with measured metrics - not doing so is what [[Mitchell Hashimoto]] describes as "[[Prompt Engineering vs. Blind Prompting|blind prompting]]". This practice emerged from the boom of [[Large Language Model|LLMs]] and is not yet established if it is a correct practice or just witchcraft.