Using both conversational and structured AI prompts can lead to more confident task delegation, which enhances academic productivity.

Well-crafted prompts are key to unlocking the full potential of large language models (LLMs) and AI assistants for academic productivity. Just as clear communication is crucial when working with people, learning to write effective prompts allows you to better leverage AI tools.

Unfortunately, just as we often find it difficult to start writing when faced with a blank page, many academics also struggle with the empty text box of a generative AI chatbot, unsure what questions to pose or instructions to give. This is a bit like having a research assistant but you’re uncertain about what they’re supposed to do. Writing prompts is a bit like delegating tasks to an assistant.

Read: Getting started with generative AI.

There are two main types of prompts:

• Conversational prompts: One effective approach is to start with conversational prompts where you have a back-and-forth dialogue to explore ideas, get feedback, and better understand its strengths and limitations. Think of it like having an initial discussion with a new research assistant to gauge their skills and knowledge areas. Conversational prompts are more open-ended, like “Suggest some potential research topics related to X” or can involve breaking down a larger task into steps through an iterative prompt-response process.
• Structured prompts: As you gain more experience, you can transition to providing more structured prompts. These include clear instructions for specific tasks – much like you would give explicit work orders to a human research assistant. For example, “Summarise the key findings and conclusions from this research paper”. Or “Create an outline for a literature review on topic X, with sections for Y and Z”. Or “Help me structure the first draft of this paper.”

If you’re less experienced with generative AI, consider starting with conversational prompts. This allows you to practice giving prompts for simple tasks until you start to feel comfortable with this kind of interaction. Conversational prompts will also help you figure out what the language model does well, and where it fails. However, using both conversational and structured AI prompts is an effective way to get the most from generative AI.

Crafting effective prompts is an iterative process of practice, experimentation, and finding what works best for your needs. Over time, you’ll develop a better intuition for scoping prompts appropriately. You’ll also learn the ideal level of specificity versus open-endedness for different use cases.