Conference Agenda

As digital transformation accelerates across low- and middle-income countries, questions remain about who participates in—and benefits from—the emerging digital economy. This paper presents a task-based empirical analysis of digital employment in Rwanda, a rapidly developing African economy pursuing a services-led model of growth. Drawing on national labour force survey data, the study applies a full-economy assessment of how digital transformation is reshaping the nature of work, not just in ICT-related fields, but across agriculture, services, industry, and the informal economy. It captures the digital content of work—analyzing how over 400 different occupations rely on digital tools, technologies, and services, and the types of digital skills and knowledge they require.

The research finds that Rwanda’s labour market is overwhelmingly characterized by low digital task intensity work, with 85 per cent of employment in roles requiring minimal digital engagement. High digital intensity jobs, by contrast, account for just 6 per cent of total employment and are spatially concentrated in a few urban districts and sectors. The findings reveal substantial disparities across demographic groups. Women, rural populations, and youth are disproportionately underrepresented in digitally intensive roles, despite policies aimed at inclusion. Educational attainment strongly predicts digital employment, though a notable share (39 per cent) of workers in high digital roles have only basic education—pointing to the importance of non-formal skilling pathways. Wage analysis shows that digital task intensity is associated with significant earnings premiums (high digital jobs earn 16 times more than low digital jobs), highlighting the labour market value of digital skills and the risks of exclusion from digital transformation. Moreover, internet usage rates among employed persons are closely tied to digital task content, suggesting that economic structure—not just infrastructure—drives digital inclusion.

By applying a granular, task-based framework to a low-income context, the study advances the evidence base on skills- and task-biased technological change and insights for inclusive digital employment policy. It concludes with recommendations to digitize traditional sectors, scale up non-formal digital skilling, and embed job content transformation at the heart of digital development strategies.

Introduction

The launch of AI tools, especially ChatGPT, represents a major turning point, fostering widespread adoption across diverse fields. While AI offers enhanced efficiency, it also poses risks to cognitive development, particularly through overreliance among youth. Relying on AI for basic tasks such as remembering, understanding, and applying knowledge hinders higher-level cognitive functions like analyzing, evaluating, and creating (expertise paradox). Excessive dependence on AI also limits Gen Z's ability to think creatively (innovation paradox). Additionally, AI disproportionately benefits experts, leaving novices at a disadvantage (equity paradox). The meta-cognitive illusion, created by AI's use, leads youth to believe they possess more skills than they truly do. Collectively, these AI paradoxes are contributing to the deskilling of youth, with potential negative impacts on the global economy, including a widening gap between high and low-skilled workers, wage inequality, economic instability, and rising crime rates. There is a lack of studies on youth deskilling due to AI overexposure, globally and particularly in India. In this context, the study aims to answer research questions: “Are AI paradoxes deskilling the youth?” and “Can AI paradox awareness reduce AI overdependence triggered deskilling?”

Methodology:

312 post-graduate university students from West Bengal, India were surveyed for this study using a structured questionnaire. The ‘Overreliance on AI’ indicator was constructed to evaluate the extent of deskilling. Simple statistical tools and regression analysis were for the analysis. In the second phase, an experimental study with 85 students, divided into treatment (made aware of the AI paradoxes and associated deskilling) and control groups, examined the impact of awareness intervention on reducing AI overdependence.

Findings:

The study signals a rising overreliance on AI, contributing to deskilling in key areas like writing, idea generation, critical thinking, research, and problem-solving which underscores the AI overdependence-induced deskilling (prevalence of innovation and expertise paradox). Many students fall into the meta-cognitive illusion, misattributing AI content as their own. Factors such as socio-economic and demographic background, tech familiarity, perceived AI effectiveness, and academic pressure, peer effect, easy accessibility etc. affect the AI overdependence. However, the study finds that raising awareness of the AI paradox can help reduce overdependence and safeguard essential skills.

Contribution:

This article advocates for the requirement and the ways of integration of specific policies (such as AI paradox awareness) to deal with AI paradoxes-induced de-skilling in the AI governance policies of different countries to address the potential wage inequality in the future and ensure social justice.

Understanding the significance of skills development and its impact on people’s lives is essential for designing effective policies that enhance labour market outcomes. While existing literature has extensively examined how changing skill demands affect workers’ vulnerabilities in high-income economies—particularly in the U.S. (e.g., Autor 2022; Atalay et al. 2020)—there remains limited knowledge about how skill supply and demand have evolved in other regions. This gap is particularly evident in emerging economies, where there is scarce evidence on which skills, or skill bundles, are most strongly associated with better employment and firm outcomes.

This paper seeks to bridge this gap by developing a comparative analysis of skill endowments and their evolution over time. A major challenge in studying skills is the availability of appropriate data. To address this, we leverage a newly constructed framework for measuring skills, enabling a systematic assessment of skill supply and demand across multiple countries. We focus on selected emerging economies with available big data from online job vacancies and applicant profiles.

First, we provide a descriptive overview of skill endowments, separately analysing the demand and supply sides. We categorize skills into 15 subcategories and assess how they are described in job postings and applicant profiles using textual analysis techniques (NLP), including word clouds that visualize the key expressions defining each skill. Second, we conduct a dynamic analysis of the evolution of skill requirements and observed skill supplies over time, utilizing time-series data to track trends across different countries.

To further understand the composition of occupational skill profiles, we construct detailed skills profiles based on applicant data from job boards, exploring whether different groups of workers systematically exhibit distinct skill offerings. We then assess skills gaps across countries where data on both supply and demand is available. Importantly, we examine whether possessing a particular skill mix (supply side) or requiring certain skill combinations (demand side) predicts future skills trends.

Finally, we investigate the relationship between various skills and labour market outcomes, including both wages and non-wage amenities, using big data methods for causal inference. By analysing cognitive, socio-emotional, and manual skills, and their 15 subcategories, we uncover significant variations in returns to skills across different economies. Our findings highlight the importance of tailored skills development policies that address country-specific challenges and opportunities, shedding light on the bundles of skills that enhance workers’ labour market transitions and improve job quality throughout the life cycle.

The world of work is undergoing profound transformations driven by rapid advancements in digital technology. Over the past decades, breakthroughs in digital tools and automation have redefined industries, reshaping labour markets worldwide. In this context, workers' ability to adapt and remain resilient has become increasingly dependent on their skills. However, while prior research has examined the employment effects of digital technologies, most studies focus on developed economies, leaving a critical gap in our understanding of these dynamics in low- and middle-income countries. Moreover, the role of skills in shaping labour market outcomes amid digital transformations remains underexplored.

This paper addresses these gaps by providing new empirical evidence on two key dimensions: (i) the impact of exposure to digital technologies on employment across countries at different income levels, and (ii) the role of specific skills in mediating these effects. Our analysis relies on the TechXposure measure developed by Prytkova et al. (2024), which quantifies global exposure to 40 emerging digital technologies. To estimate the causal impact of digital exposure on employment, we employ an instrumental variable shift-share approach, combining industry exposure scores from the TechXposure database with baseline employment shares at the first subnational level. The analysis covers 50 countries, drawing on labour force and household surveys from the ILO microdata repository.

To examine the mediating role of skills, we integrate big data from online job vacancies sourced from a job board in Uruguay. Using a skills taxonomy and empirical approach from Escudero et al. (2024), we construct subregional measures of occupational skills composition. These measures are then interacted with our subregional exposure indicators to assess how specific skills mitigate adverse employment effects or enhance positive outcomes associated with digital transformation.

Our findings provide crucial insights into the heterogeneous labour market impacts of digital exposure across different economic contexts. We find that the effects of digital transformation on employment vary significantly by country income level and pre-existing workforce composition. Moreover, certain skills—such as digital literacy, problem-solving, and adaptability—play a crucial role in buffering against job displacement while enhancing access to emerging job opportunities.

These insights contribute to policy discussions on lifelong learning and workforce development, emphasizing the need for targeted skill-building initiatives. By identifying the skill sets that enable workers to navigate digital disruptions, this research informs strategies for fostering inclusive, adaptive labour markets in an era of rapid technological change.

This research examines differences between women and men in terms of training opportunities and wages over the Covid-19 Pandemic in the UK, using lagged event analysis of longitudinal data (UKHLS). Within this approach, training and wages are viewed through an occupation and gender segregation lens, constraining the notion of individual preference, career commitment or self-selection. Building on research from Luchinskaya & Dickinson, 2019, this theory delineates between non-skills and skills-related functions of training, as well as considering the quality and participation of training to capture more multifaceted and detailed aspects of in-work skills development opportunities (Green et al., 2016). Covid-19 disruptions are grouped into two major forms: (1) employment-status related (furloughed, laid-off and reduced hours workers), and (2) the remote versus key, frontline or essential workers divide. This theory hypothesises that workers in the employment-status group suffered a ‘stagnation’ of skills, and lower wages in the long-term as a result of reduced training opportunities compared to those in regular employment able to work from home. While the second group did not see long-term wages suffer, they received less skills-related training as a result of Pandemic-related pressures. Instead, we test the hypothesis that these key workers undertook more perfunctory or non-skills-related training, mainly due to new health and safety regimes. Finally, compounding effects from existing labour market discrimination, segregation, and Pandemic home-schooling/child-care pressures, accumulate resulting in reduced skills-related training opportunities for women.

The data is treated as repeated cross-sectional individual-level data, from 2017 to 2022. The methodological approach is a quasi-experimental design, in which treatment groups (key workers and furloughed workers), are compared to control groups (those in regular employment and not in the other treatment groups), overtime, comparing pre- and post-Covid. We use firstly linear probability models to assess inequalities in terms of access in to training, and lagged linear models to estimate the long-term effect of training (or lack thereof) on wages.

Our preliminary findings are that there a key differences between key worker men and women in terms of accessing training during the Pandemic, and the long-term impact of skills and non-skills-related training on subsequent wages (scarring effects). Secondly, we also observe that these findings intersect with childcare responsibilities, in which women are more negatively affected than men.