Antonio A. Casilli’s research while affiliated with Institut Polytechnique de Paris and other places

What is this page?


This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (74)


Global Inequalities in the Production of Artificial Intelligence: A Four-Country Study on Data Work
  • Preprint
  • File available

October 2024

·

80 Reads

Antonio A. Casilli

·

·

Maxime Cornet

·

[...]

·

Labor plays a major, albeit largely unrecognized role in the development of artificial intelligence. Machine learning algorithms are predicated on data-intensive processes that rely on humans to execute repetitive and difficult-to-automate, but no less essential, tasks such as labeling images, sorting items in lists, recording voice samples, and transcribing audio files. Online platforms and networks of subcontractors recruit data workers to execute such tasks in the shadow of AI production, often in lower-income countries with long-standing traditions of informality and lessregulated labor markets. This study unveils the resulting complexities by comparing the working conditions and the profiles of data workers in Venezuela, Brazil, Madagascar, and as an example of a richer country, France. By leveraging original data collected over the years 2018-2023 via a mixed-method design, we highlight how the cross-country supply chains that link data workers to core AI production sites are reminiscent of colonial relationships, maintain historical economic dependencies, and generate inequalities that compound with those inherited from the past. The results also point to the importance of less-researched, non-English speaking countries to understand key features of the production of AI solutions at planetary scale.

Download

Digital Labor and the Inconspicuous Production of Artificial Intelligence

October 2024

·

6 Reads

Digital platforms capitalize on users' labor, often disguising essential contributions as casual activities or consumption, regardless of users' recognition of their efforts. Data annotation, content creation, and engagement with advertising are all aspects of this hidden productivity. Despite playing a crucial role in driving AI development, such tasks remain largely unrecognized and undercompensated. This chapter exposes the systemic devaluation of these activities in the digital economy, by drawing on historical theories about unrecognized labor, from housework to audience labor. This approach advocates for a broader understanding of digital labor by introducing the concept of ''inconspicuous production.'' It moves beyond the traditional notion of ''invisible work'' to highlight the hidden elements inherent in all job types, especially in light of growing automation and platform-based employment.


Making Data: The Work Behind Artificial Intelligence

September 2024

·

45 Reads

AI generates both enthusiasm and disillusionment, with promises that often go unfulfilled. It is therefore not surprising that human labor, which is its fundamental component, is also subject to these same deceptions. The development of "smart technologies" depends, at different stages, on a multitude of precarious, underpaid and invisible workers, who, dispersed globally, carry out repetitive, fragmented activities, paid per task and completed in a few seconds. These are workers who label data to train algorithms, through tasks that require the intuitive, creative and cognitive abilities of human beings, such as categorizing images, classifying advertisements, transcribing audio and video, evaluating advertisements, moderating content on social media, labeling human anatomical points of interest, digitizing documents, etc. This form of work is often referred to as "microwork". Our contribution, which documents the conditions of microwork in Brazil and offers portraits of the workers, is a step in the wider effort to overcome the current state of invisibilization. It opens up avenues for future research, with the aim of better characterizing this new form of work, tracing its changes over time in relation to the dynamics of globalization and, ideally, identifying levers for action and transitions.


Fabricar os dados: o trabalho por trás da Inteligência Artificial

September 2024

·

14 Reads

A IA gera tanto entusiasmo quanto desilusão, com promessas que muitas vezes não são cumpridas. Por isso, não é de surpreender que o trabalho humano, que constitui seu componente fundamental, também esteja sujeito a essas mesmas decepções. Na prática, para os trabalhadores, as promessas de desenvolvimento econômico e social, engendradas na origem da proposição do conceito de microtrabalho, não se materializaram. Não só as remunerações são baixas e assimétricas, mas, acima de tudo, as condições de trabalho são duras e penosas. A disponibilidade substancialmente variável de tarefas faz com que os trabalhadores migrem a todo tempo de uma plataforma para outra, o que resulta em mais tempo de trabalho não remunerado (dispendido na busca por novas tarefas, projetos e na realização de provas de admissão). Diferenças de fuso-horário com os clientes estrangeiros, majoritariamente localizados no Norte Global, forçam parte significativa dos trabalhadores a ficarem na frente de seus computadores durante a noite ou na madrugada. Trabalham em suas casas, sem encontrar clientes ou colegas. Ficam dispersos e desorganizados (com exceção de uma minoria de trabalhadores que integragem em grupos online), as plataformas os isolam e se desresponsabilizam por quaisquer danos físicos ou psicológicos provenientes de suas atividades. Como vimos, sobretudo quando se trata de moderação de conteúdos violentos, ofensivos e pornográficos, a situação se agrava ainda mais. Todavia, é notadamente porque outras pessoas não aceitam trabalhar nessas condições que as microtarefas são deixadas para grupos relativamente desfavorecidos: mães, desempregados, trabalhadores jovens com contratos precários ou que trabalham nas franjas da informalidade. Carentes de proteções sociais e trabalhistas sólidas, encontram nas plataformas uma oportunidade residual de renda. Ao mesmo tempo que fazem um trabalho central à cadeia produtiva da IA, esses trabalhadores são colocados na invisibilidade pelas plataformas.


The problem with annotation. Human labour and outsourcing between France and Madagascar

July 2023

·

153 Reads

·

26 Citations

Big Data & Society

Artificial intelligence advancements have reignited job displacement debates that focus on how the use of artificial intelligence affects labour, without considering how the production of this technology influences labour division. The generalisation of machine learning has created an increased demand for outsourced data workers. Outsourcing companies and crowdwork platforms are both used to generate, annotate, and enrich data. This data tasks are performed by workers from low-income countries, who often earn poverty wages. As with traditional outsourcing, workers must integrate complex multinational subcontracting networks. In this article, we examine how France outsources artificial intelligence-related tasks to workers in the African island nation of Madagascar. For our study, we interviewed 26 data workers, eight employees of French start-ups, and conducted secondary research on two artificial intelligence systems – a canteen checkout terminal and an algorithm to detect shoplifters in stores. The data collected allowed us to reconstruct an end-to-end artificial intelligence production value chain, revealing the need for data classification and artificial intelligence problematisation. Commercial artificial intelligence, therefore, does not displace employment by automating service jobs. Rather, by delocalising labour into the Global South, it lengthens the externalisation chain.


Microwork in Brazil. Who are the workers behind artificial intelligence?

June 2023

·

138 Reads

·

6 Citations

The report lifts the curtain on the Brazilian data workers involved in global AI production chains. They do 'micro-tasks' to generate and annotate data for machine learning, while also checking algorithmic outputs and occasionally, taking the place of failing automation. They use international digital labor platforms through which they execute on-demand tasks, mostly for AI producers located in the Global North. As platform workers, they are paid by piecework and have no long-term commitment to the buyers of their outputs. Some highlights: - Three out of five Brazilian data workers are women, while in most other previously-surveyed countries, women are a minority (one in three or less). - 9 reais (1.73 euros) per hour is the average amount earned on platforms. - There are at least 54 micro-working platforms operating in Brazil. - One third of Brazilian data workers have no other source of income, and depend on microworking platforms for subsistence. - Two out of five Brazilian data workers are (apart from this activity) unemployed, without professional activity, or in informality. In Brazil, platform microwork arises out of widespread unemployment and informalization of work. - Three out of five of data workers have completed undergraduate education, although they mostly do repetitive and unchallenging online data tasks, suggesting some form of skill mismatch. - The worst microtasks involve moderation of violent and pornographic contents on social media, as well as data training in tasks that workers may find uncomfortable or weird, such as taking pictures of dog poop in domestic environments to train data for "vacuuming robots". - Workers' main grievances are linked to uncertainty, lack of transparency, job insecurity, fatigue and lack of social interaction on platforms.


Microtrabalho no Brasil. Quem são os trabalhadores por trás da Inteligência Artificial?

O relatório abarca brasileiros envolvidos nas cadeias globais de produção de IA. Tratam-se de pessoas que realizam 'microtarefas' para gerar e anotar dados para aprendizado de máquina, ao mesmo tempo em que verificam saídas algorítmicas e, ocasionalmente, assumem o lugar de falhas na automação. São utilizadas plataformas internacionais de trabalho digital por meio das quais realizam tarefas sob demanda, principalmente para produtores de IA localizados no Norte Global. Como trabalhadores da plataforma, eles são pagos por peça e não têm compromisso de longo prazo com os compradores de seus produtos. Alguns destaques: - Três em cada cinco trabalhadores de dados brasileiros são mulheres, enquanto na maioria dos outros países pesquisados anteriormente, as mulheres são uma minoria (uma em cada três ou menos). - 9 reais (1,73 euros) por hora é o valor médio ganho nas plataformas. - Existem pelo menos 54 plataformas de microtrabalho operando no Brasil. - Um terço dos trabalhadores de dados brasileiros não tem outra fonte de renda e depende de plataformas de microtrabalho para subsistência. - Dois em cada cinco trabalhadores de dados brasileiros estão (fora dessa atividade) desempregados, sem atividade profissional ou na informalidade. No Brasil, o microtrabalho de plataforma surge do desemprego generalizado e da informalização do trabalho. - Três em cada cinco trabalhadores de dados concluíram o ensino de graduação, embora realizem principalmente tarefas de dados online repetitivas e pouco desafiadoras, sugerindo alguma forma de incompatibilidade de habilidades. - As piores microtarefas envolvem a moderação de conteúdos violentos e pornográficos nas redes sociais, bem como o treinamento de dados em tarefas que os trabalhadores podem achar desconfortáveis ou estranhas, como tirar fotos de cocô de cachorro em ambientes domésticos para treinar dados para "robôs aspiradores". - As principais queixas dos trabalhadores estão ligadas à incerteza, falta de transparência, insegurança no trabalho, cansaço e falta de interação social nas plataformas.


Number of articles published in Scopus from 2010 mentioning any of the digital platforms scored by Fairwork in the abstract (common names like “Rev”, “Prolific” or “Translated” were removed).
Source: authors’ elaboration
Research Ethics in the Age of Digital Platforms

April 2023

·

203 Reads

·

6 Citations

Science and Engineering Ethics

Scientific research is growingly increasingly reliant on "microwork" or "crowdsourcing" provided by digital platforms to collect new data. Digital platforms connect clients and workers, charging a fee for an algorithmically managed workflow based on Terms of Service agreements. Although these platforms offer a way to make a living or complement other sources of income, microworkers lack fundamental labor rights and basic safe working conditions, especially in the Global South. We ask how researchers and research institutions address the ethical issues involved in considering microworkers as "human participants." We argue that current scientific research fails to treat microworkers in the same way as in-person human participants, producing de facto a double morality: one applied to people with rights acknowledged by states and international bodies (e.g., the Helsinki Declaration), the other to guest workers of digital autocracies who have almost no rights at all. We illustrate our argument by drawing on 57 interviews conducted with microworkers in Spanish-speaking countries. Supplementary Information The online version contains supplementary material available at 10.1007/s11948-023-00437-1.


Des GAFAM aux RUM : plateformes et débrouille dans le Sud global

March 2023

·

15 Reads

·

4 Citations

Pouvoirs: Revue d'Etudes Constitutionnelles et Politiques

Les gafam sont souvent considérés comme les acteurs incontestés de l’économie numérique mondiale. Cependant, au travers de trois « ethnographies situées », nous nous concentrons sur les « rum » (Rappi, Uber et Microworkers) pour analyser les usages en ligne de livreurs mexicains, de chauffeurs en Argentine et de travailleurs du clic au Venezuela. Si, dans les pays à haut revenu, les acteurs du numérique bénéficient de la stabilité et de la protection sociale fournies par l’emploi formel, dans le contexte de l’Amérique latine, c’est au contraire en exploitant l’informalité et en réactivant des relations de « colonialité » qu’ils arrivent à extraire de la valeur. La mise en place de méthodes de débrouille et de solidarité active montre que la sphère informelle constitue un espace conflictuel où se développent des expériences d’émancipation qui peuvent aussi inspirer les luttes sociales des pays du Nord.



Citations (43)


... While the rapid recent development of generative foundation models is exciting for many potential applications (see, e.g., [1,2,3,4], etc.), important social impacts come along with rapid adoption, including worker displacement [5,6,7], use of copyrighted data for training models [8,9,10,11], energy requirements and associated climate impact [12,13], and data privacy [8,14,15,16]. To develop socially responsible foundation models, we argue for proactive consideration of such concerns across the whole research and development (R&D) lifecycle from ideation to retirement of the technology. ...

Reference:

Social Science Is Necessary for Operationalizing Socially Responsible Foundation Models
The problem with annotation. Human labour and outsourcing between France and Madagascar

Big Data & Society

... Moreover, the exploitation of low-wage laborers, or "ghost workers," who clean and categorize data for AI training, adds another ethical dimension to the use of AI in academic publishing. 11(p1), 54 Homolak highlights the dilemma of accountability in AI-generated plagiarism: "Who is to blame for plagiarism if the chatbot decides to plagiarize?" 27(p2) , emphasizing the complexity of attributing responsibility in cases where AI generates content without disclosing the source of information. 27 Huang argues that AI-driven technologies, while improving efficiencies such as text annotation and keyword extraction, often perpetuate biases embedded in the datasets they are trained on, thus raising concerns about the integrity of the scholarly outputs. ...

Microwork in Brazil. Who are the workers behind artificial intelligence?

... A escolha pela plataforma se justifica pelo fato de abarcar variadas modalidades de microtrabalho. No que concerne às questões éticas específicas das pesquisas em plataformas de microtrabalho, nos fundamentamos nas reflexões e recomendações realizadas por Molina et al. (2023). Para análise dos dados, primeiro fizemos uma caracterização das plataformas de treinamentos de dados, de maneira a compreendermos como elas se inserem em cadeias globais de suprimentos para produção da IA. ...

Research Ethics in the Age of Digital Platforms

Science and Engineering Ethics

... In fact, by monopolizing data (Zuboff, 2023), platforms exercise power and control far beyond their physical and legal perimeter, subordinating seemingly autonomous and distant organizations (Ietto-Gillies and Trentini, 2023). Finally, labour fragmentation is significantly exacerbated by platformsboth locally and on a global scale (Casilli et al., 2023) -, with relevant implications in terms of working conditions, economic vulnerability (see inter alia, Kenney and Zysman, 2020;Cirillo et al., 2023) and social conflicts (Della Porta et al., 2022). ...

Des GAFAM aux RUM : plateformes et débrouille dans le Sud global

Pouvoirs: Revue d'Etudes Constitutionnelles et Politiques

... Despite its limitations, the capacity of LPT to analyse the changing forms of work and their implications has also been demonstrated by the recent strand of literature that successfully applies LPT to a new frontier of workers' exploitation, the gig economy. This form of work organisation is centred on the intermediation and the management of labour via online platforms (Chicchi et al, 2022). Given its origin from Braverman's work, for whom the relationship between technology and power relations at work was central, it is not surprising that LPT-inspired approaches to analyse platform work have flourished (Joyce and Stuart, 2021). ...

Digital Labor and the crisis of the Wage Labor system

Sociologia del Lavoro

... However, more work is needed to grasp the complex ways in which humans augment AI across contexts and applications. Studies focusing on highly digitalized forms of work , private uses (e.g., social media and content moderation, e.g., Llansó, 2020) and domestic applications such as smart speakers (Tubaro and Casilli, 2022) show that human augmentation is both a feature and a bug of AI. Future research could systematically compare application areas and technologies, for example, embodied vs. disembodied vs. embedded AI (Glikson and Woolley, 2020), in terms of what human augmentation does to the AI and what the AI does to human augmentation (Jarrahi et al., 2023). ...

Human Listeners and Virtual Assistants: Privacy and Labor Arbitrage in the Production of Smart Technologies
  • Citing Book
  • January 2022

... For the various human actors involved, interactions with AI-based virtual assistants are increasingly becoming instantaneous acts. Furthermore, users and micro-workers are prompted to perform tasks such as data cleaning and labelling, thereby "impersonating" AI to overcome technology's shortcomings (Burrell & Fourcade, 2021;Shestakofsky, 2017;Tubaro & Casilli, 2022). By doing so, not only the algorithmic systems but also the human knowledge practices behind them become opaque in order to increase trust in their truth claims. ...

Human Listeners and Virtual Assistants: Privacy and Labor Arbitrage in the Production of Smart Technologies

... Esto confirma los hallazgos de investigaciones globales que señalan que el microtrabajo digital, aunque ofrece flexibilidad y oportunidades de ingresos, exacerba las desigualdades de género existentes. Se configura como una "tercera jornada" para las mujeres y perpetua la desvalorización del trabajo femenino (Tubaro et al., 2022). Esta dinámica se ve reforzada por la precariedad del hogar, que afecta especialmente a las mujeres con responsabilidades domésticas (Gerber, 2022). ...

Hidden inequalities: the gendered labour of women on micro-tasking platforms

Internet Policy Review

... This shift in transaction methods has emphasized the importance of identity, especially digital identity. Furthermore, the lack of a legal identity can be a critical barrier to accessing digital technologies Health and disability status 49,55,59,112,123 People living with disabilities often face greater barriers to accessing the internet, while those experiencing poor physical or mental health, or psychological distress tend to more intensely and frequently use the internet Employment status 9,[124][125][126][127] Unemployed populations have lower levels of digital literacy as baseline digital literacy is often required for many jobs. The digital world also gave rise to novel forms of employment (e.g. ...

Who Bears the Burden of a Pandemic? COVID-19 and the Transfer of Risk to Digital Platform Workers

American Behavioral Scientist

... Ghost work refers to micro-working tasks that remain largely invisible and dispersed in global (and digital) value chains (Gray and Suri, 2019;Morgan et al., 2023). Free social labor (Casilli, 2021) refers to a model of engagement not recognized as 'work' by any of the users who participate in the production process. However, in the relation between visibility and accountability, being visible means also being recognizable in the digital labor market. ...

Waiting for robots: the ever-elusive myth of automation and the global exploitation of digital labor

Sociologias