The Future of Work Working Group

This report examines the employment impacts of Generative Artificial Intelligence (GAI) in Latin America through qualitative research in Mexico, Chile, Colombia, Argentina, and Costa Rica, focusing on highly exposed sectors including call centres and customer service, graphic design and visual arts, copywriting and journalism, and software development. Drawing on more than 60 interviews with workers, business leaders, unions, and business associations, the study finds that while GAI adoption in the region remains limited, it is already reshaping tasks, productivity expectations, and skill requirements, generating new tensions around work intensity and job quality rather than clear evidence of mass job displacement. Contrary to dominant exposure-based studies that predict widespread replacement, empirical evidence points to a more complex pattern in which complementarity, task redefinition, and organisational change are more common than outright substitution, even in sectors deemed highly vulnerable. Adoption is being led primarily by larger firms and individual workers integrating tools informally, while small and medium-sized enterprises face significant capacity constraints. Training is largely individual and uncoordinated, exacerbating inequalities between workers with differing levels of digital and cultural capital and revealing the limits of reskilling as a standalone solution, particularly for women and other groups facing structural barriers. The findings highlight deep divergences between business narratives focused on efficiency and union concerns about precarisation, underscoring the risk that GAI may reinforce pre-existing labour market vulnerabilities in the absence of coordinated action. The report concludes that GAI offers a potential opportunity for more inclusive and productive labour markets in Latin America, but only if supported by deliberate social, institutional, and public policy choices that combine access to education, infrastructure, and tools with adaptive, ethical regulation that protects workers’ rights while enabling innovation.— February 25, 2026

In a context marked by rapid technological change and increasing uncertainty, artificial intelligence (AI) is reshaping work and employment at an unprecedented pace. While a growing body of research has examined these transformations through macroeconomic modelling, labour market projections, or technical and regulatory analyses, there remains a persistent need for empirically grounded, qualitative, and interdisciplinary approaches capable of capturing how AI is experienced, adopted, and negotiated in concrete work settings. This report presents the formation, objectives, and strategic trajectory of the AI@work labs network. This is an initiative designed to advance qualitative research, embedded in real-world contexts, on how AI is reshaping work and employment. By bringing together local and regional labs from around the world, the network aims to generate context-sensitive knowledge capable of informing both policy and practice at a global scale. A core ambition of the network is to give analytical visibility to perspectives that remain underrepresented in AI debates, particularly those of non-managerial workers and employees. In doing so, it contributes to an evidence-based public dialogue attentive to lived realities.— February 10, 2026

Aleah (not her real name) is an Amazon “Warehouse Associate”, working at one of the many fulfilment centres based in the UK.1 Associates, also referred to as “Warehouse Operatives” by Amazon, are the workers who receive and stow products, and pick and pack them for customers, before they are sent to another node in Amazon’s supply chain for last-mile delivery. In the words of the company, associates are “essential”; they “literally bring customers’ orders to life, every day”.2 Working in an “on task” role (also referred to as a “direct” role), Aleah and associates like her, are subject to an algorithmically determined, pace-based target, colloquially referred to as the “rate”. Aleah doesn’t really know how the “rate” is calculated. All she knows is that she needs to hit an acceptable pace to avoid coaching and potential disciplinary action, including receiving a negative ADAPT. ADAPT stands for Associate Development and Performance Tracker. According to the limited public information available and data collected for this research, ADAPT is a software used by management to track employee performance and provide positive and negative validation across a range of dimensions including productivity, quality, safety and behaviour. — May 21, 2025

The first Fairwork AI report presents the results of a case study into Sama, a data annotation company that aims to have a positive social impact, which was conducted as part of the Global Partnership on Artificial Intelligence (GPAI) “AI for Fair Work” project. Across the world, there is increasing attention paid to the precarious situation of workers that are part of the AI supply chain. The Fairwork AI ratings presented in this report, which assess the working conditions of workers at Sama in different job roles and performing various tasks, show that fairness at work is not a given, but that by pointing to shortcomings and encouraging meaningful pro-worker change, substantial improvements can be achieved. The key improvements Sama made to the working conditions of its workers through their engagement with Fairwork are the focus of this report. Concretely, as a result of engagement with Fairwork, Sama has made some 24 significant changes in the past year to their operational model and business practices, including guaranteeing the living wage for its workers, eliminating unpaid overtime and extending employment contracts.— May 21, 2025

Launched in 2021, this project is aimed at observing AI at the workplace and analyzing its impact on workers. The project methodology involves recruiting and coaching students’ communities, who conduct interviews with workers in their geographical area. The FoW WG provides the students with a common questionnaire and pays them for the use cases collected. Since this project’s inception, three generations of students have been hired from different countries, and two reports on the impacts of AI on workers have been published. The project continued using the same questionnaire as previous years but, in 2023, saw some modifications in its methodology, by incorporating a new vision focused on gig workers (this year in Mexico). The complete report is available at the link below. It presents three studies, two conducted by the Students’ Communities in Japan and Mexico, and a third from Yann Ferguson, a FoW Expert, that has included the perspective of LaborIA, a French initiative on AI at work.— May 21, 2025

The Observation Platform of AI at the workplace project was born in 2021. To understand the motivations behind this project, we can quote the first 2 sentences of this project's report in this year 2021: “To build a better future for workers collaborating with AI, to be more inclusive on various criteria such as disability, gender, ethnicity... a mandatory initial step is observation. The aim is to capture what is happening in the real context of workplaces: observe AI at the workplace, gather as diverse as possible use cases, conduct qualitative analyses of its impact in different situations, geographies, sectors, users.” And indeed, since then, three generations of students have been hired (in Europe/Canada, Japan and Mexico) and two reports have been published on the impacts that could be observed of AI on the worker's environment.— May 21, 2025

The GPAI Future of Work group has translated the OECD AI principles into concrete workplace standards to inform the practices of employers. This process of standard development initially began with a global stakeholder consultation, which generated a first version. Now, the research team have conducted two in-depth workplace studies to connect these standards to empirical data collected on the frontlines of AI system deployment. This process has allowed for a revision of the principles, conducted in association with research partner Fairwork. The new principles have a stronger focus on pay, conditions, contracts, management and representation, because these fundamental issues were consistently highlighted by research participants. A second report containing the results of the two case studies is to follow in early 2024. — May 21, 2025

n response to the expanding landscape of digital technologies and the growing complexity of AI-based software solutions, we propose a single framework for gathering the best practices related to the development of AI-based solutions. This endeavour is driven by global digitalisation and “AI-tisation”, both horizontally (i.e., covering new industries) and vertically (i.e., infiltrating existing use cases). Let us consider a case study of an autonomous intralogistic platform1 vendor based in Poland: a manufacturer of robots that usually work in warehouses. This company stands as an illustration of the transformative power of intelligent technologies: starting as the producer of Autonomous Mobile Robots (AMR), the company has evolved into a vendor of digital AI-platform supporting the design, implementation, and management of autonomous intralogistic operations for manufacturing facilities and warehouses.— May 21, 2025

Generative AI and the Future of Work remains notably absent from the global AI governance dialogue. Given the transformative potential of this technology in the workplace, this oversight suggests a significant gap, especially considering the substantial implications this technology has for workers, economies and society at large. As interest grows in the effects of Generative AI on occupations, debates centre around roles being replaced or enhanced by technology. Yet there is an incognita, the "Big Unknown", an important number of workers whose future depends on decisions yet to be made In this brief, recent articles about the topic are surveyed with special attention to the "Big Unknown". It is not a marginal number: nearly 9% of the workforce, or 281 million workers worldwide, are in this category.— May 21, 2025

The GPAI Future of Work (FoW) Working Group’s mandate and scope are to: • Conduct critical technical analysis on how the deployment of AI can affect workers and working environments as well as how workers and employers can better design the future of work. • Address how AI can be used in the workplace to empower workers, how employers and workers can prepare for the future of work, and how job quality, inclusiveness, and health & safety can be preserved or even improved. • Include a focus on the education and training needed to prepare the future workforce. The Steering Committee encourages future projects that focus on the education and training needed to prepare people for jobs of the future, as well as on how AI can be leveraged in this education and training. — May 21, 2025