Conceptualising a Framework for Effective Artificial Intelligence (AI) System Development in Terms of People-Process-Data-Technology (2PDT)

    Student thesis: Doctoral Thesis

    Abstract

    Artificial intelligence (AI) has become a ubiquitous phenomenon in recent times, with most organisations today attempting to manoeuvre their way around developing AI systems with the aim of improving the products and services they provide. However, what complicates developing AI systems is the paucity in frameworks to support organisations with effective AI system development (SD). As a result, organisations are using existing approaches which have been previously applied in earlier emerging information technology (IT) endeavours. This includes re-directing human resources with foundational knowledge and using technology which is inaugural in nature and needs to be applied and evaluated. Research highlights that this is not producing the desired results as demonstrated by the existing low success rate of AI SD projects, and several known challenges which organisations are attempting to resolve.
    The study conceptualised and examined a framework for effective AI SD aimed at improving organisational AI SD. The framework has important characteristics of agility, rigour, dynamicity, and completeness. This study also aimed to address the research question, How can organisations effectively manage AI SD?
    Therefore, this study conceptualised a framework, People-Process-Data-Technology (2PDT), for management of AI SD. It then applied a case study research design to examine the framework. The study consisted of 12 organisations in Australia. To collect data the study employed semi-structured interviews and document and website analysis. A total of 39 interviews were conducted, and 87 documents were analysed. The collected data was analysed using three rounds of thematic analysis which contributed to significant findings.
    The results contribute to determining important requirements for effective AI SD in organisations. The outcome of this research presents important requirements for organisations to adopt in order to achieve effective outcomes in AI SD. For instance, organisations need to consider appropriate skills for their people, and implement approved frameworks.
    The study contributes to theory by highlighting the importance of people, process, data, and technology in the AI domain. Given that AI technologies are evolving rapidly, they have not been tried and tested effectively, and as a result, organisations will not achieve the desired
    results from them unless they develop clear strategy to better inform technology selection decisions.
    With regard to people, the study highlights that there is a requirement for an increase in people with AI skills in the workforce. Partnerships with universities and investing in training and development have been identified as two important requirements which are essential for ensuring more people are equipped with important AI skills. With regard to process, publishing organisational AI SD findings externally was also found to be essential within the domain. This will encourage more organisations to share findings on a regular basis which will be valuable within the domain. With regard to data, the study’s findings will encourage organisations to increase their efforts to ensure better quality data is captured and maintained. With regard to technology, the study highlights the importance of risks associated with investing in inaugural technology being better understood.
    This study examined the 2PDT framework through case study research. The outcome of this examination highlighted the important role of people, process, data, and technology in the AI domain. This has contributed to a holistic view due to the characteristics of the 2PDT framework: agile, rigorous, dynamic, and complete. The agility enables seamless integration of the 2PDT into an organisation for AI SD. Following an accepted research design provides the framework’s rigour. With continuous improvement being a key element of the framework it ensures the framework is dynamic and adaptable to changing environments. A combination of case study data and existing theory makes it a framework that is complete.
    Analysis of interdependencies of the 2PDT highlights: (1) organisational culture strongly influences people, process, data, and technology, (2) people’s knowledge and experience, an agile organisational culture, and effective processes support a positive data culture, and (3) strong processes reflect organisational knowledge, experience, and culture.
    The study included cases from important fields, including education, technology, law enforcement, health care, and academia. This demonstrates that the 2PDT can be applied in different fields, industries, and environments. For example, other applicable fields include finance, sport, science, agriculture, and retail.
    The study will also empower organisations to apply the framework and identify new areas of improvement related to their context. While empirical data was collected from Australian based organisations, other organisations from the Asia Pacific region and beyond will also benefit from the findings. We believe that the key finding of the study is that the 2PDT framework will improve organisational AI SD.
    The study encountered some limitations relating to a lack of empirical results from related studies on AI frameworks, recruitment of cases and participants, and access to some documents for analysis due to organisational constraints (i.e. security, confidentiality). To minimize the impact of these limitations, the study, recruited appropriate participants from different sectors, employed an appropriate methodology for data collection and analysis to ensure the empirical analysis have been appropriately explored and discussed.
    This study concludes with directions for future work in the Asia Pacific region, and globally, in different organisations. There is opportunity for future work to study additional organisations, and to evaluate the 2PDT framework in other geographical locations and organisations. This will expand upon the findings of this study, contribute to the analysis and effectiveness of the 2PDT framework, and also to knowledge in the AI domain. There is also opportunity for future work to focus on application of the 2PDT framework in organisations to examine the results produced.
    Date of Award2024
    Original languageEnglish
    Awarding Institution
    • Faculty of Science & Technology
    SupervisorHamed Sarbazhosseini (Supervisor), Masoud Mohammadian (Supervisor) & Sarbazhosseini (Supervisor)

    Cite this

    '