It is not recommended to simply ask an AI tool to suggest articles on your topic, as AI tools can and do simply make references up!
However, there are things you can use AI tools for that can help in the process of your research. For example:
For more information on searching for resources see our guide.
AI tools can be particularly helpful when it comes to refining and improving your academic writing. For example, you can:
With text based AI the answer that you receive is based on the text that you input. Prompt engineering is spending time considering what information you want to input to improve the generated output. This is similar to how you would use specific terms when finding information with online databases.
See below for advice on how you can improve your instructions/questions through prompt engineering.
When giving prompts try to avoid vague or ambiguous language. Use specific terms and try to make it as clear as possible what you are asking.
Example:
Non specific question - "Tell me about climate change"
This original question is broad and would give information that would lack any detail.
Revised question - "Can you provide an overview of the main factors contributing to climate change, including the role of greenhouse gases and human activities and their impact on global temperatures over the past century".
The revised question adds specifics to guide the AI to give more detailed information.
Include context where possible. This could include timeframes, who the audience of the text is, how long you want the answer to be, what format (bullet point list, letter, email, article) the text should be in, relevant topics, your knowledge level etc.
If the initial response did not give the information that you were looking for, you can ask follow-up questions to refine the responses. Text based AI use the previous questions and information given to inform their response to a follow-up question.
For example:
If the information is vague you could ask "can you provide more details about [x]".
At the start of your prompt you can assign the AI a role for your questions. The AI will then respond in the capacity of that role. This role can include answering as a specific person, as a job title and as someone with certain level of experience in an area.
Example: "You are a communications expert who communicates in a friendly and engaging way. Write me a draft email about..."
If an initial response does not work, you can ask the AI to "regenerate" the response. This will give you a newly generated answer that may be different to the original. This is quite effective when asking the AI to create something for example, metaphors, examples or when asking AI to explain concepts to you.
The best way to learn prompt engineering is to 'play around' by trying different things. Ask questions in different ways, test the impact of adding pieces of context and assigning different roles. To gain an understanding of how AI works you may find it easier to start on topics that are not academic
Activity - Test using different phrasing or instructions on the same topic
Using a text based generative AI trial asking similar questions with different wording or instructions. Pick a topic, ask a simple question and then build on the question with context, roles and specific language. Note down your reflections on the differences in responses.
