The development of artificial intelligence (AI) has made various industries eager to realise and obtain the benefits of AI. There is an increasing amount of research surrounding AI, most of which is centred on the development of new AI algorithms and techniques, thereby, however, ignoring an increasing set of practical problems related to AI mode lifecycle management, examples include versioning control of the data and models, difficulties in model deployment, transparency, model reproducibility, fairness and balance in data, and ethics. In contrast to this extensive list, research on the AI model lifecycle management is limited, and there is currently no comprehensive study. To address this gap, we researched the life cycle management of AI models. The research consisted of two parts. First, we conducted a systematic mapping study, which consists of the classification and counting of published papers in this field. By summarising the current situation of this field through quantitative and qualitative research, we obtained an overview, identified research gaps, and provided suggestions for future research. We then selected one of the specific themes, AI democratisation. Using this theme, we researched relevant literature, highlighted the importance of model documentation and the research gap, carried out research on the existing model documentation framework, improved the current framework, and introduced a solution for promoting AI democratisation. Specifically, we proposed a tool to automatically generate model documentation. Finally, we conducted a user study on the tool as a means of gathering suggestions for future improvements. Purpose: Improve the life cycle management of AI from a theoretical and practical perspective through conducting a comprehensive study of the life cycle management of artificial intelligence models; a solution to a specific topic/research gap (i,e., the democratisation of AI). Method: We carried out systematic mapping research on the life cycle management of AI models. Regarding AI democratisation, we researched literature pertaining to the democratisation of existing AI, compared with existing solutions, and improved AI from the perspective of generating model documents. Results and Conclusion: The systemic map was obtained by categorising and counting related publications, improving the existing model document framework, and proposing a solution to automatically generate model documents, thereby improving AI democratisation.