Rules to hack Prompts

Rules to hack Prompts


Many quickly become disappointed with language models like ChatGPT.
A person writes a request like, "Give me a plan on how to make a million," and in return, they receive a list of trivial tips.
"It doesn't work like that."
"Imagine asking such a question to an intelligent person. They will provide clear instructions for action."
No, they will ask a series of clarifying questions: about your abilities, available capital, free time, knowledge, and willingness to take risks. And in the end, they will give an equally banal answer that you have to work hard and spend less.
Language models are just a tool, a database with the ability to emulate natural language. But it's an excellent tool; you just have to learn how to communicate with it.
How can I formulate questions in ChatGPT?
  1. copy the prompt
  1. paste in chatGPT
  1. If there is a placeholder, usually in this format [placeholder], replace it once you will have copied the prompt inside ChatGPT.

6 Rules Prompts:

1) First, create a persona on behalf of whom the model will communicate with you. This is done as follows:
"Imagine being a talented business person who has created many companies": this is how we suggest what information you should use to communicate with us.
2) Next, we set the task:
"Create for me a lis of easy-to-implement business ideas using which I can quickly earn money."
3) Action plan (and limitations) for #ChatGPT:
"I am not interested in local ideas that can only be implemented in specific countries or ideas that require large resources to be implemented."
4) Provide more context:
"Give preference to original ideas that can be implemented online using #ChatGPT or other language models."
5) Setting the output:
"Divide the result into paragraphs to make them easy to read."
6) Communication rule:
"When you are done, write end. Give me the opportunity to ask clarifying questions, indicating the number of a specific idea.”
Therefore, the ideal prompt for the language model looks like this: