Finding articles with AI: faster access to relevant studies and insights

—

© 𝘈𝘯𝘥𝘳𝘦𝘢𝘴 𝘖𝘩𝘳𝘦𝘯𝘴𝘤𝘩𝘢𝘭𝘭

In a recent project, we supported research into AI solutions used across different business processes. More specifically, the task was to identify articles and studies in which IT consultants presented detailed use cases for AI applications.

A manual Google search produced some initial results, but these were mostly too general and often lacked the desired industry focus. We therefore turned to ChatGPT in a second step, where the added value became much clearer:

  • The findings were automatically sorted according to the predefined process categories and the authors of the use cases.
  • This provided a clear overview and made it much easier to navigate the material from the outset.
  • ChatGPT also listed the relevant AI applications separately for each category in text form, making it easier to use the information directly in subsequent analysis.

Our conclusion: Using language models to search for studies, publications and similar material has considerable potential for research tasks. Prompting is especially important. The clearer the requirements – such as industries, categories or sources – and the instructions for the output, the more relevant and structured the results are likely to be.

Our recommendation for this type of research: Invest time in developing and wording the prompt. In our case, we named selected IT consulting firms as examples, clearly defined the process categories and narrowed down the relevant industries. This made it possible to conduct a more targeted search and produced results that were useful from the outset.

By Maria Kleiner