How can generative AI help in different parts of research? An experiment study on smart cities’ definitions and characteristics

Artificial intelligence (AI) engines, such as ChatGPT, InferKit, and DeepAI, are very popular today and new AI engines, such as Google Bard, Chinchilla AI DeepMind, and GPT-4, constantly emerge. However, question remains how these new data management tools can assist scholars in improving the research design and implementation. In an attempt to answer this question, we focus on one particular research field – definition and identification of smart cities (SCs), – and compare the answers provided by different AI engines with the answers given in a sequence of research papers, prepared without the use of AI and recently published by these authors. In particular, the following aspects of the original studies were re-analysed here using the AI input: a) problem definition; b) summary of current knowledge; c) identification of unknowns; d) research strategy, and e) recommendations for research and practice. As the study reveals, the recommendations of AI engines are, at times, inconsistent and data sources cited are often inaccurate. However, as such engines scan multiple open sources and retrieve relevant information, they can help to bridge gaps in the summary of background studies and streamline the research design, by supplementing missing or overlooked information.