Skip to Main Content

Generative AI for Learning and Research

What is a prompt?

Prompt, also known as prompt engineering, is the process of designing and refining a notion in text for intelligent computers to interpret and generate a requested output. In simple words, prompt is the conversation you have to interact with the AI.

Traditionally, the understanding of computer and programming language are essential for creating a prompt. With GenAI, especially the implementation of large language model and machine learning, we can now communicate with the computer in natural language to generate any desired outputs.

Constructing a prompt

To have the GenAI tool produce output that is relevant, useful and as desired, a well-crafted prompt is needed to avoid any irrelevant response.

To build an effective conversation, below are some tips and suggestions on how to create a prompt:

  1. Be specific - know what you are looking for
    Instead of using ambiguous or abstract wordings, make use of clear and precise terms to explain your request. Try to avoid the inclusion of unnecessary details or information to ensure the scope is focused on what you are trying to achieve. You may add criteria to set boundaries, such as providing specific instructions by saying what is expected and what is not expected or stating word limits or formats of the outputs (e.g. present the content in table format with five bullet points). 
     
  2. Provide context - set the stage
    Like creating a fictional story, you can set up a context that enables the generation of product to be more focused and relevant. You can create a character for the machine and include the information in the prompt for it to act on, such as “you are a research assistant in a recognized university.”. Involvement of the targeting audience would also help to specify the output, like “you are preparing a seminar on information literacy for year one business school students”. Of course, you can provide as much detail as possible; still, remember to be precise and exclude irrelevant or unnecessary information which may hinder the results.
     
  3. Giving examples - allowing more details and references for the machine to analyze on
    Providing examples, or even real-life cases, may facilitate the analysis of the machine and make it more practical. This could serve as a reference point for AI to enhance the outputs. For instance, “describe the uses of AI tools for student’s research; for example, brainstorm and polish writing”. 
     
  4. Iteration - practice makes perfect with trial and error
    There may be times when the generated outputs are not as expected or not satisfying. In this case, you are encouraged to give feedback to the machine for it to modify the results. You can always ask follow-up questions for further elaboration that based on previous prompts, like “provide more details or examples”, or “how...”
    If there are cases where the outputs have become irrelevant or not useful, you may consider revising your prompt and starting a new conversation or thread by clearing the previous contents.

 

References: