The Unseen Laborers of Artificial Intelligence: Surviving the Crisis in Venezuela

EL PAÍS offers the América Futura section open for its daily and global informative contribution on sustainable development. If you want to support our journalism, subscribe here.

The economic crisis in Venezuela forced Oskarina Fuentes seven years ago to become an invisible Artificial Intelligence (AI) worker. Her role is to label data to improve the performance of internet robots in exchange for the minimum money to survive. “They are more than all searches,” says the 33-year-old woman, dedicated to gathering information from companies and people, selecting the best answer to a search criteria, moderating content so that atrocious scenes stop circulating on the network, among other endless “tasks,” which are adding pennies on the dollar in her Appen account.

The Role of Data Taggers in AI Systems

Australian virtual platform Appen compiles data for tech giants like Microsoft, Amazon, and Google to hone their AI systems, with the help of contributors from more than 170 countries, who register on its website and select the tasks they want to perform.

Data taggers or annotators like Sources provide information to computational models so they can make decisions, from improving web searches to enabling more complex algorithms like those of a self-driving car to work. “The system is watching and learning from what they do,” explains Dr. Alberto Delgado, an AI expert at the National University of Colombia.

Surviving the Economic Crisis in Venezuela

Behind the scenes of this billion-dollar industry, the payment that Fuentes has left ranges between 200 and 300 dollars a month, which is close to the minimum wage in Colombia (209 dollars), a country to which she migrated in 2019 with her mother. The deep economic crisis that Venezuela has been going through for a decade forced many to look for alternative methods to survive. And data labeling platforms, which do not require any special qualification, were presented as a viable option to alleviate hunger.

“Slaves of Latin America”

Fuentes, a petroleum engineering graduate, suffers from diabetes and precarious health that has prevented her from practicing her profession and accessing another job. Hundreds of Venezuelans with whom she talks on the Telegram social network about her experiences in Appen find no other way to survive either.

The platform, valued at about 500 million dollars according to an Australian media, sets the remuneration of its collaborators seeking to “exceed the minimum wage in the region,” which is not difficult to achieve in Venezuela, where this indicator is 5.4 dollars a month, after more than five years of hyperinflation.

“With a lot of effort, I manage to earn about 200 dollars a month,” says a worker who prefers not to reveal her name for fear of reprisals from the company. Her earnings are the result of her work on Appen and other similar websites such as Toloka, Hive Micro, Testable Minds, and Paidera. The money she earns from them is barely enough for her food, her husband, and her two children, who have no other income. “The work is enslaving and poorly paid,” says Rodriguez, who clings to these virtual tasks from the Venezuelan city of Cabimas.

Rodrigo Sircello, who does this work from Maracaibo, says that he and his partner registered in 2016 with the promise of generating a good income. “My wife constantly received emails from Appen (…) In her advertising, it was read that it was a remote job and great profits could be made,” says the 57-year-old man.

However, in 2023, he and his family are struggling to make ends meet in the absence of homework. “Since the beginning of this year, it has been difficult to achieve the minimum charge, which is 10 dollars a week,” says the father of the family, who uses all the monthly money from his retirement as a librarian to pay for the internet at his home and be able to use Appen. Regardless of how many years they have been registered on the platform, collaborators do not have any formal link with the company, nor do they have the guarantee of receiving tasks. In addition, their work often does not match the hours of the area in which they are; therefore, given the need, Venezuelans submit to work at any time.

“I have sleep problems,” emphasizes the worker from Cabimas, who lives with the computer “on 24 hours a day,” in case she receives a notification of a task at dawn that will help her make ends meet. When problems arise on the platform, the woman claims that Appen is slow to respond to complaints or not at all. “They don’t answer my tickets,” says the housewife. Constant power cuts make her work more difficult. Faced with the treatment of his collaborators, Appen said in an email to América Futura that he “deeply values his workers because they represent the factory of the companies where they operate,” but he refrained from answering specific questions about the conditions of Venezuelans on the platform.

The Colonialism of Artificial Intelligence

Fuentes’ story was revealed in April 2022, within a series of reports from the Massachusetts Institute of Technology (MIT) magazine in which they spoke of the “colonialism of artificial intelligence.” In them, through several cases, they portrayed the power of large companies in the industry over collaborators from developing countries, who work in precarious conditions. The examples reinforce “the idea that AI is creating a new colonial world order,” according to the magazine.

From that publication, the name of Fuentes was quoted in the media around the world, in a way in which she does not entirely agree. “I don’t feel like a slave to either Appen or the AI,” says the young woman. “We are slaves to the Latin American system,” clarifies Fuentes, who believes that living in a low-income region is what determines the lack of guarantees.

Earlier this year, the magazine Time alerted to a similar case, in which the OpenIA company subcontracted people in Kenya taking advantage of the impoverished economy of the African country to filter toxic texts from ChatGPT for a payment of two dollars an hour.

The Need for Regulation

A lover of anime and animals, Oskarina Fuentes emphasizes that she speaks out loud about her experience so that Appen listens to her collaborators who are “trained and hard-working people.” “We want them to recognize our efforts and take us into account for greater opportunities,” says the young woman from her residence in a town in Antioquia, Colombia. Dr. Alberto Delgado, an expert in AI, assures that the problems of these collaborators reside in the lack of control in that market. “AI touches human beings. For this reason, ethical principles must be applied that pave the way for the regulation of the industry,” highlights the university professor.

Last month, the European Union and the United States announced the advancement of a draft common “code of conduct” on AI, which would be applied on a voluntary basis in the future. In a manual of recommendations on the subject published in 2021, UNESCO warned that due attention should be paid to lower-middle-income countries “which are more exposed and vulnerable to the possibility of abuse of a dominant position in the market”.

Meanwhile, without active regulations or guarantees but eager to pay the bills, Fuentes and her colleagues in Venezuela want “Appen to keep working” and clamor for “more tasks” to appear, while they wait 24 hours next to their computers, with the anxiety typical of a country that is collapsing.