AI models have been boomed recently, and many dull work and assignments are distributed to large-scale models, saving people much time. However, we only know their charm and appreciate the convenience, but no systematic principles and rules are applied for better use experience and effective response. In this article, we first concentrate on normal mistakes when using ChatGPT, and then I will give you a universal formula for using artificial intelligence.
Common mistakes
- Provide insufficient background: Most beginners would make this mistake. It is expected because everyone who has just used AI has no idea of what the interaction will be. In a word, they don’t know the AI functions modules and what kinds of assistance they can be provided. It is understandable because when we first come across dialogue assistants, we just reckon them as human beings’ helpers and they would just react like humans. Remember the first time you use Siri?
- Unset roles before asking questions: Many users expect ChatGPT to do translation or text2text generate work. But they sometimes ignore the various styles and tones in articles and papers. For example, a 10-year-old student wants ChatGPT to modify his assignment paper and just fix some grammar mistakes or enhance the expression while ChatGPT then generates a lengthy and jumbled response, which is just like it’s written by a professional academic writer.
- Too eager for the outcome: Some users want their problems to be quickly solved. They jump too many steps to get the final answer. A typical question can be like this: Write an article titled ‘How to use ChatGPT correctly’. There is no issue with the question itself. But just throw it directly to AI, you can only get a vague and maybe run-of-the-mill article. Because one paper includes an abstract, discussion, questions, and conclusions. Each part of them is consistent and has a relation to do with others. AI can only generate an article with no highlights and is quite worthless. People always want AI to provide ‘direct’ and ‘straight’ help and answers, but the more straight it is, the less value it has.
A universal formula
Now see this formula: Role + Background + Mission + Steps = Success
To make ChatGPT effective and productive, just follow the formula mentioned above. Let me explain it.
- Role: Tell your model who it is. If you want it to solve a medical problem, just let it be a doctor or professional nurse. There are jobs in the world, you should give your models one before they work for you.
- Background: Tell your model in what circumstances it solves problems and provides advice. Also, tell it your expectations. Remember the more background it receives, the more accurate answers it will generate.
- Mission: Tell your model what it should do directly. After setting roles and providing enough backgrounds, you can just give ChatGPT tasks. It is easy because everyone has their problems, and the key is to describe the mission as clearly as possible.
- Steps: Tell your model what to do step by step in detail. We use the ‘writing article’ case that is mentioned. The right method to let ChatGPT write an article is to tell it how to write each paragraph or part and what exactly the content each part should contain. For example, you can say: Write an article titled ‘How to use ChatGPT correctly’. Firstly you should write some issues while using ChatGPT and refer to some common mistakes. Then give each issue a reliable solution and evaluate its feasibility. Finally, give me a formula that can be applied to ChatGPT users.
Follow these simple rules, I think you can truly harness the skills of using ChatGPT!