How Madagascar’s low-paid workers fuel French tech’s AI ambitions

How Madagascar's low-paid workers fuel French tech's AI ambitions

Basic drawing of a data annotation company made with AI Midjourney. Credit: Généré par MidJourney CC-BY-4.0, CC BY-NC

ONE January 2023 survey of Time magazine revealed that Kenyan workers who are paid less than US$2 an hour are given the job of trying to ensure that the data used to train the ChatGPT AI platform is free of discriminatory content. .

AI models need to be trained, with huge chunks of important data as input, in order for them to learn to recognize and interact with the human environment. These inputs need to be collected, sorted, verified, and formatted. Such time-consuming and underrated tasks are often outsourced by technology company for a precarious contingent of workers, often based in the Southern Hemisphere.

This data work comes in different forms, depending on the purpose of the final algorithm. For example, it might involve sketching people in images captured on video cameras to teach algorithms how to recognize people. Or one might be checking the output of an automated invoice processing tool and manually correcting errors to help the computer do its job.

To explore the identities of these data workers, their roles and working conditions, and enrich the debate surrounding AI regulation, we set up an investigation. taken between Paris and Antananarivo, the capital of Madagascar.

Our research also shows the reality of AI, the French style: on the one hand, French tech companies depend on the hosting and processing power of the Big Five (Google, Apple, Facebook, Amazon, and Amazon). Microsoft); on the other hand, data tasks performed by workers in former French colonies, especially Madagascar, confirm long-standing outsourcing trends. Incidentally, there was a study comparing the field of technology with mining and textiles.

A study on the globalization of AI

Our research project kicked off in Paris in March 2021. We first set out to find out what French AIs are involved in data processing and what processes are involved. applied to ensure the creation of high-quality data sets to train computer models. We conducted interviews with 30 founders and employees working at 22 companies in Paris in the AI ​​ecosystem. One quick finding emerged from this initial poll—most data work is outsourced to Malagasy contractors.

For the second part of the study, conducted at first remotely, then on site in Antananarivo, we interviewed 147 workers, managers and directors at ten Malagasy companies. At the same time, we sent a questionnaire to 296 data officers based in Madagascar.

Precarious work for young intellectuals in the city

Our initial requests suggest that AI data workers are part of a much broader IT services sector, from call center staff to web content moderators, to copywriters search engine optimization (SEO).

Responding to the questionnaire showed that the majority of workers working in the industry are men (68%), young (87% under 34 years old), live in urban areas and have an education level (75% have experienced, achieved have or have had at least some college education). When work takes place in the formal economy, rather than the black or gray economy, respondents are often full-time employees. The minimal protections offered by Madagascar, in contrast to French labor law, the ignorance of workers about their rights, the weakness of trade unions and Worker representation in Malagasy companies adds to the precariousness of their position. Most of them earn between 96 and 126 euros a month, with a huge gap between their salaries and those of group supervisors: who also tend to be Madagascans, working in the country, but taking home 8 to 10 times more.

Factory workers find themselves at the end of an extremely long processing chain, which partly explains the very low wages even by Malagasy standards. The AI ​​production line involves three different parties: data storage/processing power services provided by Big Five technology companies, French companies selling AI models, and companies providing AI models. provides data annotation services provided by Madagascar workers. Its each level cuts.

Data-contesting companies are often heavily dependent on their French clients, who manage an outsourced workforce in a near-direct manner, forcing middle managers to work for the good. of startups in Paris. The dominance of these roles by foreigners – either employed by client companies in France, or expats working in Antananarivo – presents a serious impediment to career growth. of workers, who remain shamefully stuck at the bottom of the value chain.

Profit from post-colonial France-Madagascar links

The AI ​​sector benefits from a specific policy—a “tax-free zone” created in 1989 for the textile industry. Since the early 1990s, French businesses have set up satellites in Madagascar, especially for the digital publishing industry. Special zones, equivalent to those found in many other developing countries, attract investment by offering very attractive tax exemptions.

Today, of the 48 businesses offering digital services in tax-free zones, only 9 are owned by Madagascar, compared with 26 that are owned by the French. Besides the situation with formally established companies, the sector has developed a tiered subcontracting operation, with gray economy businesses and entrepreneurs at the bottom of the pecking order, mistreated and must act when there are workforce shortages elsewhere in the sector.

Along with cheap labor, this outsourcing industry benefits from a highly educated workforce—most of whom have attended college and speak the French fluently they learned in school, online, and online. online or in the classes of the Academy of Francais. The organization later introduced the French language and culture, founded in 1883was originally intended to extend imperial power through language to the colonists.

This scenario is consistent with what researcher Jan Padios noted “colonial recovery.” Colonies that formerly had linguistic and cultural ties to once-dominant countries now offer them business services.

Make AI employees visible to better understand how they operate

Behind the recent boom in commercialized AI projects in the Global North, a growing number of data workers are discovered. The recent controversy around “smart security cameras” at the Paris Olympics mainly focused on the ethics of comprehensive surveillance. The important component of human labor needs to be taken into account in training AI models, especially since it raises new questions about working conditions and the right to a private life.

To make the role of these workers clear is to ask probing questions about globalized production chains. These are more familiar in the manufacturing industry, but are also a feature of the digital realm. These workers are essential to the operation of our digital infrastructure—they are invisible cogs in our digital lives.

It also shows the impact of their work on AI models. Part of the algorithmic bias lies in the nature of the way the data is conducted, although the fact of this is largely kept secret by artificial intelligence company. Therefore, a truly ethical AI must establish moral standards because Working conditions in the field of AI.

Provided by

This post was reposted from Conversation under Creative Commons license. Read original article.Conversation

quote: How low-wage workers in Madagascar fuel French tech’s AI ambitions (2023, March 30) get March 30, 2023 from 2023-03-low-paid-workers-madagascar-power-french. html

This document is the subject for the collection of authors. Other than any fair dealing for private learning or research purposes, no part may be reproduced without written permission. The content provided is for informational purposes only.


News7F: Update the world's latest breaking news online of the day, breaking news, politics, society today, international mainstream news .Updated news 24/7: Entertainment, the World everyday world. Hot news, images, video clips that are updated quickly and reliably

Related Articles

Back to top button