ľ¹Ï¸£ÀûÓ°ÊÓ

Generative AI (GenAI) offers potential benefits for researchers and support staff. Keep in mind that GenAI is not a substitute for expertise and human insight. It should therefore be used as a supplementary tool, when relevant and appropriate. Do not overly rely on GenAI in your work.

Benefits and use cases

A few examples of benefits and use cases of GenAI in research (support) include:

Drafting and improving text

GenAI can quickly generate first drafts for sections of a document, such as an article or proposal. Since writing a first draft can be a barrier, GenAI can provide a starting point to build on and save researchers and research support staff valuable time. It can also help to polish text or make it more concise (for example to fit a word limit).

Data analysis and pattern recognition

GenAI excels at identifying complex patterns in large datasets, which can reveal insights that might be missed by human researchers. Always be very critical about the results as they may be false or biased.

Cross-linguistics inclusiveness

A good example of GenAI use is using automatic translation to overcome language barriers for non-native speakers (for example, translating emails). Since research papers and proposals tend to be written in English, GenAI is particularly helpful for non-native English speakers.

Task Automation for research projects

GenAI can help drafting progress reports, meeting minutes, or create summaries of meetings or communication chains.

Summarize extensive/complex information

GenAI can summarize lengthy or complex documents, making it easier to extract key insights and communicate important points to colleagues. For example, funding support staff can  use GenAI to summarize and distill dense funding calls or eligibility criteria, ensuring that researchers are presented only with the information most relevant to them.

Experimental design optimization

AI models can help design more efficient experiments by suggesting optimal parameters, sample sizes, and methodologies based on previous studies.

Literature review assistance

GenAI may facilitate literature reviews by synthesising or summarising literature, but users should always fact-check the results. GenAI may not be suited to identify relevant literature or summarise it correctly. Outcomes depend on what the model was trained on, and/or whether the model is able to generate based on live web search. More reliable tools exist; an example is ASReview, developed at UU.

Code Generation

Increasingly, researchers find GenAI models useful to either generate programming code based on a linguistic prompt, to translate code from one language into the other, or to get useful hints on code continuation within their programming environment, as well as creating documentation and assisting with de-bugging of code. Code-specific models also exist.