Few shot learning is the easiest way to get better outputs from LLMs. Find out how and when to use it, it's current limitations, and tons of examples.