Prompting is a skill you can train. The more you experiment with different formulations, the better you learn what works. There are a number of do’s and don’ts that can help you communicate more effectively with AI tools.
Below is a list of elements you can consider including in your prompt:
Description:
Clearly describe your task or instruction.
Example(s):
"Summarize this article in 150 words."
Description:
Simulate a persona for tone and style.
Example(s):
Description:
Outline steps and/or the sequence of the result.
Example(s):
"First discuss the benefits of AI and then conclude with possible next steps. Describe the next steps in a bullet-point format."
Description:
Provide sufficient context and/or constraints.
Example(s):
"Use the following sources to answer my questions. [Links to your sources and/or PDF files]."
Description:
Define goal and target audience.
Example(s):
"Rewrite this paragraph for a first-year pedagogy undergraduate student (university level). The purpose of this paragraph is to inform them about […]."
Description:
Give instructions about the output.
Example(s):
Good prompts are therefore clear, specific, and aligned with the goal. That’s why it’s important to first be aware of what your goal is. Based on your goal, you can determine which elements from the table above are needed to make your prompt as complete as possible. So instead of simply asking, “Summarize this article,” it’s better to ask, “Summarize this article in 200 words for a second-year psychology student, with emphasis on the article’s conclusion.”
You can also ask AI what information is still missing to carry out the task as effectively as possible. Additionally, various prompt libraries are available that contain example prompts:
Prompting is therefore more than just ‘asking something’. It’s an iterative process in which you adjust your prompt based on the generated responses. In that sense, you can also compare it to learning how to argue: through language, you try to guide a system that has no intentions or goals of its own, but does learn statistically which word sequence is most logical.
Curious to see a demonstration of good versus poor prompting? Then watch the video below:
(The video is in Dutch, but you can use the subtitles).