Large language models have shifted the human–artificial intelligence (AI) interaction paradigm in business processes. The purpose of this exploratory study is to examine key opportunities and challenges that human resources (HR) staff face when applying prompt engineering techniques during their interactions with large language models to perform operational and strategic HR tasks.
This study uses creative prompting techniques to test and use generative AI guided by HR managers’ queries. Methodology includes zero-shot, few-shot and chain-of-thought prompting. These prompting techniques enhance the model’s adaptability and accuracy by leveraging different levels of HR task guidance and cognitive processing.
The study found that operational HR-related tasks are performed well, whereas strategic HR practices are not. This operational-level HR task capability limits the use of generative AI for HR-related tasks by hindering its ability to capture and fully represent HR domain knowledge. The study provides some suggestions to solve such limitations.
This study has taken “HR practices” as an illustrative example to demonstrate the practical implications of using generative AI.
This study relies on ChatGPT’s HR-related domain knowledge demonstrated in its responses to prompts supplied by HR managers. Prompt quality may typically suffer due to the modeller’s lack of domain knowledge. Thus, the approach uses three prompt engineering methods designed to extract domain knowledge of HR using generative AI. The study identifies a gap that emerges when humans interact with AI to perform strategic HR tasks. This gap shows the need for GPT modellers to improve GPT efficiency for strategic operations.
