How AI is cre­at­ing jobs in In­dia, not just steal­ing them

With In­dia emerg­ing as a hub for data la­belling, a new work­force is help­ing com­pa­nies in US, Europe perfect their ma­chine learn­ing models

The Times of India (New Delhi edition) - - Times Sport - [email protected]

Five years ago, Hy­der­abad res­i­dent Tu­lasi Mathi was forced to quit her job as a maths teacher due to health is­sues and the birth of her two children. But today, the 29-year-old does data la­belling and makes up to Rs 15,000 a month. The money isn’t much but it’s more than she made as a teacher, and enough to pay her kids’ school fees and her own ex­penses.

She chanced on data la­belling work through a YouTube video in 2017. Today, she scans videos and marks and la­bels ob­jects en­coun­tered by self-driv­ing cars. Her out­put is used to train ar­ti­fi­cial in­tel­li­gence al­go­rithms pow­er­ing such cars. All Mathi knows is that it makes her life eas­ier. “I can work from home and don’t have to choose be­tween work and fam­ily,” she says.

Mathi is one of the face­less workers help­ing com­pa­nies in the US and Europe perfect their ma­chine learn­ing models. For in­stance, if you’re try­ing to get a driver­less car to cor­rectly iden­tify a stop sign, you need to feed that al­go­rithm thou­sands of images cor­rectly la­belled as stop signs. Sharmila Gupta of Gur­gaon-based AI Touch likens data la­belling to training a new­born. “Any AI model re­quires la­belled data to get trained. This is like teach­ing a small child mul­ti­ple times till they un­der­stand.” It’s a job that only hu­mans can do and since it is quite labour in­ten­sive, it is be­ing out­sourced to coun­tries with cheap labour like In­dia, Malaysia, the Philip­pines and Kenya.

There is a new AI work­force, says Ajay Shah of HR com­pany TeamLease Ser­vices. From an op­por­tu­nity point of view, there are about a lakh jobs posted on var­i­ous por­tals cur­rently.”

Mary Gray, a re­searcher and au­thor of the book ‘Ghost Work: How to Stop Sil­i­con Val­ley from Build­ing a New Global Un­der­class’, says there are two main rea­sons why com­pa­nies are turn­ing to In­dia for data-la­belling and an­no­ta­tion. “It has a work­force trained in English as a first lan­guage and an in­ter­net in­fra­struc­ture created dur­ing the first out­sourc­ing boom that re­lied heav­ily on In­dia as an off­shore labour mar­ket,” says Gray. There is also a grow­ing de­mand for data-la­belling ser­vices that are “lo­calised”— both lin­guis­ti­cally and cul­tur­ally rel­e­vant to In­dia — and this work can’t be done by workers in the United States.

Play­ment, a crowd­sourced mar­ket­place that trains an­no­ta­tors from scratch, where Hy­der­abad’s Mathi works, has 25,000 an­no­ta­tors be­tween the age of 18 and 30 years work­ing re­motely across In­dia, and its co-founder Sid­dharth Mall claims that any­one with a lap­top and basic English skills can start work­ing. “Ev­ery­body talks about how AI will make peo­ple lose their jobs, but there are also new kinds of jobs be­ing created,” he says. These young­sters earn any­where be­tween Rs 20,000 and Rs 30,000 a month.

It’s not a for­tune but it’s help­ing

many an­no­ta­tors — most of t hem stay-at-home moms, fresh grad­u­ates and even stu­dents, such as 21-year-old Shiekha Ma­hara from Nal­go­nda, Telangana — get by. Ma­hara, who re­cently com­pleted a BTech de­gree, be­gan look­ing for on­line work op­por­tu­ni­ties to help out with her fam­ily’s fi­nances. She has earned Rs 1.3 lakh so far while do­ing oc­ca­sional projects for Play­ment over the last year and a half.

Un­like Play­ment which pays peo­ple on a con­trac­tual ba­sis, iMerit, a data an­no­ta­tion com­pany with of­fices in In­dia and the US and data la­belling cen­tres in Ranchi, Shil­long, Vizag and Kolkata, has 2,500 peo­ple on their rolls. What they have in com­mon is an over­whelm­ingly young work­force. At iMerit, the aver­age age of em­ploy­ees is just 24. Jai Natara­jan, VP, mar­ket­ing and strate­gic busi­ness de­vel­op­ment, says that nearly 80% of their em­ploy­ees come from un­der­priv­i­leged back­grounds, while 50% are women. “Our em­ploy­ees are po­si­tioned for the fu­ture be­cause they un­der­stand that they have to learn new things, that noth­ing stays still,” says Natara­jan. iMerit’s em­ploy­ees do data la­belling for drones in the agri­cul­ture sec­tor, med­i­cal im­agery such as MRI scans, e-com­merce, and sports an­a­lyt­ics.

Mu­jeeb Ko­lasseri, a high school dropout from Man­narkkad in Ker­ala, founded his own data la­belling com­pany In­folks in 2015 af­ter learn­ing the work on­line. Today, the com­pany em­ploys nearly 250 peo­ple, nearly half of them from poor fam­i­lies in Ko­lasseri’s vil­lage. New em­ploy­ees get trained on im­age an­no­ta­tion tools for two months. “Nearly 80% are fresh­ers. With proper training, any­one can work on im­age an­no­ta­tion with­out any tech­ni­cal knowl­edge — you just need to be a quick learner,” says Ko­lasseri, who was forced to quit his studies be­cause of his fam­ily’s fi­nan­cial prob­lems.

Ji­ten­dra Ku­mar, 27, would agree. Six months ago, Ku­mar, who used to drive a four-wheeler for wed­dings and par­ties in his home­town Etawah, Ut­tar Pradesh, found a data la­belling job with Gur­gaon-based firm AI Touch. “Now, I get a salary on time, work in an of­fice and can spend some time with my fam­ily as well,” says Ku­mar.

Ku­mar’s col­league Satyam Barth­wal, a Chi­nese in­ter­preter, was hired af­ter the com­pany got work from a Chi­nese AI com­pany. “I hadn’t even heard of data la­belling be­fore I got the call,” says Barth­wal, who sees the job as a way to earn money till he ful­fils his dream of be­com­ing a singer. “The work is easy — we just need to read la­bels.”

But as ma­chine learn­ing evolves, will it make the work of data la­belling re­dun­dant in the fu­ture? “Since we started in 2013, the pre­ci­sion, nu­ance and so­phis­ti­ca­tion required has gone up. Some­times we need do­main ex­perts. But even then, you need hu­mans to re­view, au­dit and keep track of re­sults. There is go­ing to be a role in AI for hu­mans for a long time,” adds Natara­jan of iMerit.

LA­BEL LABOUR: (Clockwise) Data la­belling workers in Noida, Ker­ala and Kolkata

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.