ChatGPT's primary purpose as a language model is to perceive and respond to user inputs. However, the formulation of these inputs can have a significant impact on the quality of the following conversation. This post will provide ten suggestions for improving the effectiveness of prompting in ChatGPT.

Back to top

1. Be clear and concise

Do: "Can you explain the difference between MLA and APA citation styles in academic writing?"

Don't: "I'm confused about writing styles. Can you help?"

Back to top

2. Use specific examples

Do: "Can you provide examples of how machine learning is being used in the healthcare industry?"

Don't: "What are some applications of machine learning?"

Back to top

3. Be mindful of context

Do: "Can you explain the difference between weather and climate, and how they relate to each other?"

Don't: "What is climate change?"

Back to top

4. Avoid leading questions

Do: "What are the arguments for and against the death penalty?"

Don't: "Don't you think the death penalty is inhumane?"

Back to top

5. Use proper grammar and spelling

Do: "Can you explain the difference between 'there', 'their', and 'they're' in English grammar?"

Don't: "Can you tell me about there, their, and they're?"

Back to top

6. Use clear language

Do: "What is the impact of social media on mental health?"

Don't: "What are the repercussions of digital platforms on the psychological well-being of people?"

Back to top

7. Ask follow-up questions

Do: "Can you explain more about the concept of 'intersectionality' and how it relates to social justice?"

Don't: "What is intersectionality?"

Back to top

8. Be patient

Do: "Can you explain the difference between socialism and communism?"

Don't: "Why is socialism better than communism?"

Back to top

9. Use diverse prompts

Do: "Can you explain the basic principles of Buddhism?"

Don't: "What is religion?"

Back to top

10. Provide feedback

Do: "Thank you for explaining that! However, I was hoping for a more specific answer to my question. Can you clarify?"

Don't: "Your answer was wrong. Try again."