An ordinary day in our “digital life” looks something like this: on a laptop, computer or phone, we ask a search engine to see what is being shown tonight at a movie theater or the addresses of nearby restaurants. Many relevant recommendations emerge. Or, our favorite music app suggested a playlist that matches our taste. Social media feeds are mostly devoid of offensive content. None of this would be possible without the hidden army of thousands of workers from all over the world.
These workers are not “from Silicon Valley” but the so-called “microworkers” – to perform tasks that machines cannot. These jobs have a bad reputation – they are poorly paid and involve hours of hard work. For many people, though, this is the perfect job. The so-called Microworkers generate data by transcribing, cleaning, correcting, and categorizing content. They help to supply data to machine learning algorithms. These algorithms form the basis of artificial intelligence. They do this by adding a human element to the data.
They can draw border frames around road shots, teach automated cars how trees look, road obstacles and people on the move. They also tag content with feelings. Sentiment analysis algorithms can learn what a “sad” song sounds like, or whether some text or word is “troubling.”
Amazon owner Jeff Bazos called crowdsourced micro-work “artificial intelligence” when it launched the world’s first platform of its kind in 2005. It’s called Amazon’s Mechanical Turk. It was named after an 18th century chess player. The device pretended to be players competing with the machine, while in reality they were playing against a real person, moreover a real chess master, who was hidden inside the machine.
Amazon was the first to use it to clean millions of duplicate product pages. Computers couldn’t see the subtle differences in the pages, but people did.
Because one man would not be able to perform this huge task, they broke it into small, discrete, repetitive tasks. These tasks can in theory be completed by human workers in just a few seconds from anywhere in the world. This model was opened to all companies through a website that served as an intermediary between companies looking for those tasks.
Microworkers from all parts of the world who perform them are paid on an errand
It is difficult to estimate the global crowdsources of micro-workers because companies do not publish any official figures other than their registered users.
The most famous is Amazon’s Mechanical Turk. It is estimated that tens of thousands of people work at MTurk every month. 2,000 to 2,500 of them are active at all times, at any time – this number was calculated by Panos Ipeirotis of New York University.
This study found that the majority of MTurk workers are from the US and India. This is because the platform has for many years restricted access to non-U.S. Workers. There are now many more microradio platforms around the world.
The International Labor Organization (ILO) has tried to map the situation. It surveyed 3,500 micro-workers in 75 countries from all over the globe.
They found that the average age of the respondents was 33 years. A third were women. That figure is different in developing countries, where women represent a fifth.
They were also educated – less than 18 percent of those surveyed had a high school diploma or less. A quarter studied at university, 20 percent had a post-graduate degree. More than half had a degree in science or technology, 23 percent in engineering, and 22 percent in information technology.
The micro-city can be a lifeline for people who do not have access to traditional employment or sources of income, especially in countries in crisis.
Mishel Munoz lives in Venezuela. It has been online for the past two years. A failed economy and rampant hyper-inflation make Venezuela’s micro-city the best possible option for Mishel to make money. “I used to have an office. But, unfortunately, I had to close it because of migration and because people do not have enough money to afford a dentist, they have to worry about food, education and other things. I do a lot better this way than as a dentist, “she explains.
Jahia Ayub Ahmed is a Syrian living at Darashakran Refugee Camp in Erbil (Iraq) after escaping the war.Here he learned how to be a micro-worker. “You can work remotely and generate money. Applying for a job here is very difficult, you can’t just google the job so it allows you to work without having to apply and submit a CV, ”he explains.
He learned how to perform extremely precise image annotation tasks (putting bounding frames around images or drawing segmentation masks). “I’m very excited to be learning about machine learning and artificial intelligence,” he says. . The tasks are directed mainly to developed countries, explains Allen Nisu, of the Receptive Lion.
“Most online co-sourcing services restrict Iraqis from being able to apply, but even if someone miraculously succeeds, there is no way to cash in on the work being done, since all transactions occur through a third-party online banking system, such as is Pipel, so they have to negotiate through special agreements with companies. ”
It’s harder for people who work alone from home.
Rafael Perez lives in Venezuela. He makes a living through microradiation. He describes one such situation. He says he is unable to do most of the tasks required. He adds that he lost money many times to unscrupulous tenants who did not pay him. “I remember once making $ 180 in 15 days of work. That’s a lot of money, but they never paid me. I sent them emails, called them … but you don’t really have anyone to complain about. ”
The microradio is often criticized for being poorly paid. An ILO poll found that workers earned, on average, $ 4.43 an hour. Salaries vary from case to case. Fees vary by region. US workers earn an average of $ 4.70 (still below the guaranteed minimum wage there), while workers in Africa are paid $ 1.33. Fees do not take into account that workers spend an average of 20 minutes on unpaid activities for each hours of paid work.This involves searching for assignments, taking tests to show that they are qualified for the task, and answering test questions put together by the employer for quality control.
“I spend all day or most of the day looking for tasks. When one comes in, I have to sit down to do it right away, “says Rafael. They also spend time avoiding fraudsters.
Mishel says she initially fell for fraudulent sites. Now she takes care to check them out first and give the sites a try before they even start working for them. The way these workers are paid is also not classic. Workers who are paid electronically have to turn it into cash. They use pir-to-pir platforms where brokers charge a commission to convert electronic credit into cash.Rafael says he can earn eight to ten dollars a day on a good day. For this new one can buy a small carton of eggs, a kilo of corn flour and a kilo of beans.
“You will not live very well, but you will eat well,” he says.
He focuses on what he calls “bleak” but easy tasks of finding information on the web and tagging parts of sentences. Meanwhile, Michel does tasks with translating, drawing borderlines and analyzing sentiment. She tells the BBC about one good day when she made $ 80. She used that money to buy a smartphone she now uses in her microradio. She further thinks that there should be more “support and recognition” for microradio. The impression is that platforms make a lot of money from their work and “what they pay us, even here in Venezuela, is very small.”
Paola Tubaro, who has studied the practice of microarrays, thinks that this type of work is not transient but of structural importance for the development of new technologies (artificial intelligence).
“Even if machines learn to distinguish cats from dogs, for example, and do not need more examples, you will still need to supply them with new details to recognize traffic signs in other languages or countries.” As these technologies expand , so will the need for people to fill them with data, “so it’s not something that is temporary,” she says.
There is currently no state regulation for microradio platforms. The platforms themselves determine the working conditions. It also criticized the lack of support for those who download moderated content. They are exposed to disturbing content, which they have to deal with in the end.
The ILO calls for better regulation of the sector in order to meet basic conditions (minimum wage and greater transparency of payments).
For the time being, Tubaro says, “the microradio is invisible to many and attracts very little public attention.” “” It can’t be fair if these people aren’t paid enough or don’t have some sort of social protection. If people work under the same conditions as in 19th century factories, our society will not accept it as something that can be right. ”
Cover photo: http://www.blog.mturk.com
Photos: http://www.turkitron.com, http://www.cloudacademy.com