Mapping workers skills across AI supply chains across geographies

This paper critically examines the multifaceted role of worker skills within global AI supply chains, building on existing literature to explore the contributions of both data and algorithmic labour. We map the diverse skill requirements across various AI applications, including large language models (LLMs), autonomous vehicles, medical image diagnostics, virtual assistants, content moderation, security systems, among others. A key focus is to understand the specific tasks performed and the corresponding skills required at different stages of AI development, with particular attention to how these dynamics vary across geographies, especially within the Global South, and their implications for local labour markets and their economies.

The primary objective of this paper is to analyse the relationship between workers' existing skills and their formal education, and the practical demands of their skills within the AI supply chain, and to understand how these tasks facilitate or hinder career progression in local labour markets. Specifically, we examine how workers' self-perceived expertise aligns with their daily tasks, assessing the implications for skill matching and training needs by analysing online job vacancies and interviewing workers performing these tasks in business process outsourcing (BPO) companies involved in AI development. We also explore the extent to which these tasks performed relate to workers' formal education and fields of study, and how this alignment influences their professional development. Furthermore, the research probes whether workers perceive AI-related tasks as beneficial for their future career growth, identifying factors that contribute to or detract from this perception. Finally, we examine how these experiences, particularly regarding skill acquisition, influence workers' ability to secure future employment in the private or public sectors, by providing an analysis of how participation in AI supply chain tasks shapes or contributes to overall employability and career trajectories.

To achieve these objectives, the research employs a mixed-methods incorporating a quantitative analysis of job vacancies and interviews with workers involved in AI development. Specifically, we analyse job vacancies from online databases to better understand the skills, qualifications, and competencies sought by companies hiring for roles in BPO companies. These findings are then compared with data collected through interviews with workers holding positions with similar job titles, identified via LinkedIn and other career-oriented social media platforms, with the aim to understand whether workers' expectations align with the realities of their work content, skill gaps, working conditions as well as professional growth.