Adaptive self-regulating guidelines for the use of Generative Artificial Intelligence in business customer experience

Abstract

This research interrogates the Generative Artificial Intelligence moral dilemma in South African business customer experiences. Hereto, businesses contend with striking the right balance between capitalising on the competitive gains of GenAI while averting the inherent risks. Moreover, to exacerbate this dichotomy, businesses have to confront the lag in regulation amidst the fast-paced GenAI developments capable of autonomous updates in real time. To grapple with the GenAI regulation lag and moral dilemma holistically, the research explores the role of adaptive self-regulation. Adaptive self-regulation by businesses was recommended as systemic government-led policy development requires much of the state’s resources and takes time to implement, oversee and manage change. To this end, a qualitative research design was adopted to garner in-depth business GenAI regulation insights through the lived experiences of the participants. This corresponded with the interpretivist philosophical underpinnings of the research. Data was collected from technology experts who are involved in the sphere of technology in South Africa. These experts also work with technology, and their job titles and experiences are described in the participant characteristics table. Hereon, these are referred to as technology professionals in South Africa (n=21). Collectively, they had between 5 and 30 years of experience. To enrich the study, an exploratory design was used to cross-examine the implications of the perspectives of the research participants. The findings were scientifically analysed using ATLAS.Ti and the themes were in congruence with the research objectives. These included the benefits of GenAI, the gaps of GenAI and the enabling factors of self-regulating GenAI. To this end, the insights drawn from the research participants met the objectives of the research. From the perspectives of the research participants, the GenAI benefits are ubiquitous. They recognised speed, saving time, automation, customisation, elimination of mundane repetitive tasks, large data analysis and more. At the same time, they noted with concern the ethical and moral inherent gaps, such as a lack of human touch (warmth and empathy), hallucinations, security threats, algorithmic biases and others. The research participants were also concerned about the businesses’ position on the safety of customers’ personal information. These concerns illuminate the relevance of adaptive self-regulatory guidelines. The findings recognised that the guidelines should enhance transparency, safety, and diversity and limit the proliferation of social injustices. Even more concerning is that in the literature review findings, most businesses’ GenAI policies have remained outdated on a static webpage and are not as current as the GenAI technology. This research gap points toward the contribution of this research in advancing agile and adaptive GenAI governance. Similarly, another research gap highlighted by the research participants was the limited knowledge on how to adaptively self-regulate GenAI. Thus, this research contributes to the academic body of literature by exploring the steps toward developing adaptive self-regulation. This exploration was undergirded by the theoretical underpinning of Bandura’s Social Cognitive Theory (SCT) of self-regulation. There was convergence between the findings in theory and practice. These findings show the significance of human oversight, self-monitoring of performance, exercising judgement, and self-reaction and self-efficacy. Research participants also highlighted the role of proactive thinking as opposed to reacting to gaps. This included benchmarking, evaluations, testing, data masking, secure prompt engineering and several other proactive regulatory measures. The role of a multidisciplinary approach was also identified as a limitation by the participants. Thus, they recognised the impact of a human-centred approach that encompasses all facets and disciplines. Similarly, South Africa’s National AI Policy advocates for a human-centred approach to the development of these technologies. Hence, this holistic approach of aligning theory with the research findings culminated in the development of 10 steps for the development of adaptive self-regulation guidelines.