Skip to Main Content

Online Study Skills Hub: Finding Information with AI

Competencies essential for academic and professional success

Finding Information with AI

While GenAI tools can be used to retrieve information and enhance your learning in all the areas noted above, they do not replace library resources – such as UR Library Search and specialist academic databases – or key types of academic publications, such as textbooks, monographs and journal articles.  

Your lecturers still expect you to support your arguments with high-quality information sources and engage with the academic publications prescribed in your Resource Lists and provided by the library. As such, GenAI tools should be used to supplement and enhance your use of library resources.

We recommend that you develop your research skills through our Tutorial Series and our Finding Resources Step-by-Step Guide, and that you explore the specialist library resources listed on your Subject Resource Guide.   

 

Enhance Your Searching 

Generative AI tools can be used to retrieve information and develop and enhance literature searching strategies. You may find AI-generated summaries useful in developing your knowledge of new concepts – which can be useful as you begin a research project or assignment – and you may find that AI can provide useful suggestions in developing a search strategy to apply to other search tools. 

When using GenAI to develop search strategies, follow the principles outlined below for writing effective prompts and using GenAI critically.

In this example, we’ve asked ChatGPT to ‘find some journal articles on coaching and mentoring’.  

Screenshot of a ChatGPT interaction

ChatGPT’s response suggests that our prompt could be improved in terms of its specificity (or explicitness in the CLEAR model).  

Screenshot of a ChatGPT adaptive prompt

Our response is adaptive – it refines our initial prompt by specifying several factors. It also develops the prompt from one requesting that ChatGPT does the work of finding journal articles, to one giving useful suggestions for how we can systematically search specialist library resources (databases) for relevant journal articles. This results in a more relevant, comprehensive and useful response that will allow us to access appropriate academic sources. 

Screenshot of a ChatGPT suggestion for a search strategy

In summary, ChatGPT is not an authoritative source of information on our topic, nor does it necessarily have access to the most relevant, high-quality and up-to-date academic literature. However, it can help us develop a strategy for how and where to find that literature. 

You can also enhance your searching with Lean Library, a browser extension that integrates with Google, Google Scholar and other search engines, to detect when content is available to you via UR Library subscriptions.

Write Effective Prompts

Write Effective Prompts

Screenshot of a ChatGPT prompt entitled 'what is a prompt in the context of using ChatGPT'

GenAI tools work by being given instructions or questions – known as prompts – and generating text (or other types of content, such as images and code) in response. Just like when searching in academic databases or search engines, the usefulness of GenAI output results is affected by the quality of the prompts provided.  

Screenshot of a ChatGPT prompt entitled what makes an effective prompt when using ChatGPT

Formulating effective prompts is known as ‘prompt engineering’. Try following the CLEAR Framework (Lo, 2023) when engineering your prompts: 

Concise 

  • Use clear, concise, and specific language in your prompts. 

  • Remove unnecessary language to allow AI to focus on the most important aspects of the prompt. 

Logical 

  • Provide a logical flow and order of ideas in your prompt – ensure your prompt details the progression of steps you would like AI to follow. 

  • Structure information in order of importance in your prompts. 

Explicit 

  • Explicit prompts with precise instructions will help the AI tool produce relevant, helpful results.  

Adaptive 

  • Be flexible and adjust your prompts to improve results. 

  • Think strategically about how your prompts could be improved and experiment with adjusting them. 

  • Do not be discouraged if a prompt does not produce useful results; rephrase it and ask AI again. 

Reflective 

  • Critically evaluate AI-generated responses and reflect on them in relation to your expectations. 

For more prompt engineering guidance, see our eLearning Team’s Top Tips for Using AI Effectively, Learn Prompting’s Prompt Engineering Guide, and OpenAI’s Prompt Engineering Best Practices.

Screenshot of a ChatGPT prompt entitled what makes a bad prompt

Bad prompts are those that generate unclear, inaccurate, or unhelpful responses. This can happen when prompts are vague, ambiguous, overly complex, lack sufficient context, or are made up of excessive jargon, technical terminology, imprecise or incorrect language.  

 

Reference

Lo, L. S. (2023) ‘The CLEAR path: A framework for enhancing information literacy through prompt engineering’, The Journal of Academic Librarianship, 49(4), article number 102720. Available at: https://doi.org/10.1016/j.acalib.2023.102720 

Limitations & Reliability – Using GenAI Critically

Limitations and Reliability of ChatGPT

Screenshot of a ChatGPT prompt entitled what are ChatGPT's limitations

AI-generated content isn’t always reliable! Responses may be outdated, biased, or simply wrong. Here are some considerations to bear in mind when engaging with GenAI tools: 

  • Most GenAI tools are not search tools and are not designed to support literature searching for academic contexts. Their access to academic sources is limited and they are known to invent sources altogether (a process known as ‘hallucinating’). 

  • GenAI tools don’t usually disclose the sources of their information, making it difficult to assess the reliability and accuracy of responses. Even when a source is cited, as noted above, it may not actually exist.  

  • High-quality academic research often involves a systematic search for information on a topic; it’s important that readers know how a researcher searched for their information and are able to reproduce that search to further research in the field. AI responses are not reproducible and, therefore, are not suited to systematic searching.  

  • As we’ve seen above, AI tools can be good at summarising information and providing introductions to new topics, but they do not usually provide the level of detail required at university. 

  • GenAI tools are biased! For more information on how AI-generated output can perpetuate the biases inherent in the data on which they’re trained, see the University of East London’s AI guide

For these reasons among others, it’s essential that you use AI critically. To do so you can:  

  • Request sources or evidence from the GenAI tool. For example, you can ask, “Can you provide evidence to support your answer?” This can encourage the provision of more reliable and verifiable information.  

  • Use multiple prompts to refine your request and response received. Try rephrasing or providing additional prompts to get a more accurate or comprehensive answer. Iterative refinement can lead to better results. 

  • Ask the GenAI tool to explain its reasoning or provide a step-by-step explanation. This can help uncover any potential flaws or biases in the generated response. 

  • Always fact-check AI-generated information. It is your responsibility to ensure that the information you use in your work is accurate. Corroborating AI-generated information through trusted academic sources will help you identify misinformation and back up your work with references to high-quality sources. 

  • Address AI-generated biases by engaging with multiple perspectives and sources.