You’re correct that GPT-3 and later can accomplish such tasks with what is known as a zero-shot approach. In-context learning with few-shot exemplars are neede

You’re correct that GPT-3 and later can accomplish such tasks with what is known as a zero-shot approach. In-context learning with few-shot exemplars are needed for things like encouraging responses that adhere to a certain JSON data type.