Many well-educated people have to make ends meet with low-paying click jobs. For a few cents, for example, they label photos, transcribe audio recordings or tag videos. This data is then used to train the algorithms of online shops, language assistants or self-driving cars. The insatiable demand for such services has led to a great need for cheap Clickworkers, as reported by MIT Technology Review magazine in its current 6/2022 issue.
Demand for Clickworkers exploded in Venezuela
In the past five years, crisis-ridden Venezuela has become a hotspot for this work. It fell into the worst economic catastrophe of any peacetime country in 50 years just as the demand for Clickworkers was exploding. Scores of well-educated people with Internet access signed up for crowdworking platforms to survive. This gave companies some of the cheapest labor ever. “There is an enormous power imbalance,” says crowdworking researcher Julian Posada, a doctoral student at the University of Toronto. “The platforms decide how something is done.” Much of it is reminiscent of the colonial era.
Drivers were, of all things, car giants of the old school, such as Volkswagen and BMW. Autonomous cars need millions to billions of training data – for example in the form of hours of videos in which Clickworkers mark all road markings, vehicles, pedestrians, trees or garbage cans.
What exactly does artificial intelligence have to do with power, domination and colonialism? We address this question in the current issue. The new issue is available from August 18th. in stores and from 17.8. easy to order in the heise shop. Highlights from the magazine:
For a long time, Amazon’s “Mechanical Turk” was the dominant agent for such low-wage labor. But as a platform for generalists, it could not guarantee sufficient quality for safety-critical applications. In the early 2010s, a new generation of specialized crowdworking platforms such as Appen, Remotask, Hive Micro and Spare5 emerged.
The Clickworkers’ payment depends, among other things, on how high the minimum wage is in their home country, what experience they already have and how quickly and precisely they work. According to Remotask, clickworkers in Venezuela earn an average of around 90 US cents an hour. TR has created its own account to verify the statements. After two hours of work, including a tutorial, the earnings were eleven cents.
Minimum amounts before withdrawal is possible
On some platforms, the money can only be withdrawn after the minimum amount of 10 dollars has been reached. This can take weeks, because the tasks set are often not solvable – for example because the instructions are incomplete or because videos cannot be played due to technical bugs. Users also complain about missing payments for weeks or months. This is proven by screenshots available to TR.
“It is a mixture of poverty and good infrastructure that makes such phenomena possible,” says crowdworking researcher Florian Alexander Schmidt, professor at the HTW Dresden. “If the crises continue, it is very likely that another country will take over from Venezuela.” In fact, some platforms are already specifically looking for even cheaper labor in Asia, Latin America, Africa and the Arab world. “I think they know that people have problems here,” says a Clickworker from Kenya.
Venezuelan Oskarina Fuentes Anaya is one of the people who depend on such jobs. She studied oil and gas engineering, but there have been no jobs in the oil industry since the economic downturn. In order not to miss any lucrative jobs, she now only goes for walks at the weekend and also leaves her computer running at night so that she can be woken up if necessary.
Anyone who protests against working conditions runs the risk of being banned. Fuentes has also been banned from Appen for allegedly “dishonest answers”. The support confirmed that it was an administrative error. Still, it took months for her account to be recovered. Despite everything, Fuentes is infinitely grateful: “I survived thanks to this platform. Other platforms stopped paying, but Appen was always there.”
Read MIT Technology Review:
To home page