Dinesh Raj, who lives in Salem, Tamil Nadu, cherishes his account on microwork website Amazon Mechanical Turk even though competition for data annotation tasks on the crowdsourced platform is high and the salary is weak.

The 30-something, who has an engineering degree, struggles to find a well-paying job, and relies on the platform for much of his income, which can fluctuate daily.

READ ALSO | A new global standard for AI ethics

“I work at night, when there are more jobs from American customers,” said Mr. Raj, who has done work on Amazon MTurk for about four years.

“Out of 10 tasks I do, only two can be approved, so I have to do more tasks to earn $10-30 a day. But it’s still better than nothing,” said Mr. Raj, who sometimes praises his ID card to members of a Facebook group of Indian workers on Amazon MTurk.

The explosive growth of artificial intelligence (AI) drives the need for vast training datasets, which are served by millions of workers tagging text, images, video and audio for everything from assistants from voice recognition to facial recognition and 3D image recognition for autonomous vehicles.

India accounts for about a third of online freelancers globally, according to the International Labor Organization, with developing countries accounting for about two-thirds of the total remote workforce.

Lax labor regulations and low wages are the norm, even as workers handle the most tedious and grueling jobs with few legal protections, tech experts say.

“There is some rhetoric about data annotation creating new opportunities for people who need flexibility or need to work from home. But workers are at the bottom of the AI ​​value chain,” said said Urvashi Aneja, director of the Digital Futures Lab, an Indian. research collective.

“A lot of this work is very precarious…with ChatGPT and other generative AIs, it’s also emotionally taxing. The issue of content moderation is going to get more complex, and we’re going to see more low-cost workers. salary taken in there,” she said.

Outsourcing

The launch of San Francisco-based OpenAI’s ChatGPT chatbot to the public in November sparked a global frenzy, with more than a million users downloading it in the first week.

Since then, the AI-powered tool has been incorporated into education, marketing, customer service queries, as well as online research and content creation.

With the excitement came the revelation that OpenAI had outsourced data annotation to workers in Kenya, who were paid less than $2 an hour to tag content that included hate speech and images. violence and sexual abuse, according to a Time magazine survey.

But they are not the only ones.

Chicago-based Hugo regularly hires annotators from abroad, who are usually college graduates and whose first language is English, French or Portuguese, according to his website.

One such worker – known as an assessor – in Jibowu, Nigeria, said that although the work was not difficult, “the number of tasks can sometimes be unimaginably insane, so you would find yourself work beyond the contractual working hours on certain days”.

“Overtime is not paid unless explicitly requested,” he told the Thomson Reuters Foundation on condition of anonymity to avoid retaliation.

Every minute of an eight-hour workday is tracked “to the nearest second”, the evaluator said, and he could be penalized if he failed to do so by constantly interacting with the screen with his mouse. If he didn’t keep up with his time, he would have to start his hours over and lose his pay for any time that showed him as unavailable, he said.

“It’s extremely exhausting. You have little or no time for anything else,” he said. Even when he completed a task, he had to keep clicking on the page until his eight hours were up, he added.

The assessor’s payslip said he was paid 127,500 naira ($274) for a month’s work in January, about $1.50 an hour, about four times the minimum wage national fee of 30,000 naira per month.

A spokesperson for Hugo said the company’s teams “operate on a maximum workday of eight hours…When the client requests overtime, we offer it to our team members, who can voluntarily working them fully remunerated on pre-agreed bonuses”.

“We have young men, women, stay-at-home moms and recent graduates on our team, all of whom need the flexibility to keep busy and pursue careers,” they said in an emailed statement.

In the Philippines – long an outsourcing destination for its young English-speaking population – freelance data annotation on platforms such as Upwork has become “very competitive”, said John Anthony Abayari, in the central region’s Bulacan province. of Luzon.

Abayari, 25, has worked as a freelance data annotator for local and foreign companies since 2019 when he “really needed a job”, starting with a monthly salary of 12,000 pesos ($218).

Last year he started working as a freelance video annotator and now earns around 40,000 pesos a month.

The work can be “difficult because it is time consuming and tedious”, he said. “For the money, it’s always worth it. But if I’m lucky enough to have another job, I’ll take it.

workers

India is one of the world’s largest markets for data annotation work, with up to 1 million full-time and part-time workers by 2030, according to the industry body NASSCOM computer.

Currently, around 50,000 data annotators are freelancers on platforms such as Amazon MTurk and Clickworker, while around 20,000 work full-time at third-party companies, many of which are based in smaller cities and towns.

These companies offer workers training, fixed hours and monthly salaries of around $200 to $300, which are particularly attractive to women, said Muzammil Hussain, founder of Tika Data, which serves global enterprises.

“The work itself is simple, so customers don’t pay much. It got me thinking: How can I reduce costs? By moving to smaller towns and villages where costs are lower “, did he declare.

“The money will always be low as it is low skilled work. But ₹15,000-25,000 is a very good salary in a village, where there are few other opportunities for women, so they are happy” , he added.

However, self-employed women in India can even struggle to access microwork platforms, often paying large sums on the gray market for an account and settling for the lowest-paying and most tedious tasks, according to a report by 2021 from the IT for Change research group, which studied Amazon MTurk.

Less than a quarter of Indian workers on the platform are women, he said. Amazon MTurk did not respond to a request for comment.

“The fact that this work is almost entirely digital and can be distributed can actually allow women to participate. It is considered respectable work that women can do without leaving their homes, and it is ‘clean and safe’ work. “”, said Sarayu Natarajan. , founder of the Aapti Institute, a digital think tank.

“However, women face specific barriers: lower levels of access to digital technologies – phones and laptops – and difficulties in accessing credit finance for the purchase of devices,” he said. she stated.

For Anu K., the path to a job in data annotation was easier.

Living in Mannarkkad in Kerala, she had few job options despite her master’s degree and was a stay-at-home mom. So when she heard about Infolks, a data annotation company near her that was hiring women, she was thrilled.

Ms. Anu was hired, and after two months of training, she started working full-time, earning about $220 a month.

“I find the job interesting and my family is very happy because I work from 9 a.m. to 5 p.m. and no night shifts,” she said. “Mannarkkad is a small town, so there aren’t many options for women. Being in IT was my dream, and it’s the best option.”