NEW DELHI: When a video showing factory workers sewing clothes with small cameras on their heads went viral this month, it sent chills across social media. It also reignited a familiar fear among Indian labor activists: that people are already training their own replacements in real time.
The footage that in early April made the rounds on social networks and headlines in the local Indian press shows a garment production hall, where a few dozen people are sitting in rows, working on sewing machines.
At least 10 of them are wearing head-mounted camera gear, as a person recording the video walks by them, saying in Hindi: “Everybody has put on the camera. Even the timer is on. Everyone is wearing the camera.” One of the workers responds: “Alright, sir.”
Both of them speak with a Hindi accent that is common in the eastern Indian state of Bihar, from where most workers in India’s textile hubs in Noida and Gurugram come.
While the owners of local factories that Arab News reached out to denied using cameras to capture workflow data or having knowledge of such practices at their facilities, those working in the artificial intelligence industry say it is not new.
Senthil Nayagam, founder of Muonium AI, told Arab News that IT companies and some specialized machine learning firms that create datasets have been hiring employees on a temporary basis to do so with small sports or DJI cameras.
But what is seen in the footage is not such a gig.
“It’s not an IT company,” Nayagam said. “This hardware seems to be custom hardware. It’s not standard, market available ... The company which is getting it done is evaluating it (with) robots, and they want to see if the training data can help them reduce costs.”
Whatever it means to “reduce costs,” to labor activists it is already worrying in the Indian context. According to data from the All India Trade Union Congress, out of the country’s 670 million workforce, only 80 million people enjoy work protections.
For Amarjeet Kaur, AITUC secretary-general, using workers to build automation tools or gather data they do not benefit from only adds to how companies are already exploiting them and “bypassing” labor laws.
“Our workforce is already in a very, very dangerous zone ... With the entry of AI, naturally, many will be losing jobs,” Kaur said.
“We are not saying the new technology should not come, but you have to protect the livelihood of everybody and the dignity of everybody ... We have failed because nobody will be able to stop technology or stop AI, and especially when it is in the control of corporates.”
What may be lost is also the hope that AI would only relieve people from unsafe jobs.
Sucheta De, social and political activist, worries that in India it will make livelihoods even more insecure instead of offering workers more free time for leisure and doing what they want.
“Rather than getting some free time off from labor, they will be devoid of any kind of employment,” she said.
“The reality is — in a country like India — that until and unless the government provides some kind of job security to the huge majority of informal workers, AI is only here to displace people from their livelihoods.”
And those earning from the process may ultimately not even be contributing to the Indian economy.
Subimal Bhattacharjee, policy adviser specializing in digital technology, equals training AI models on factory workers with “extracting human skill” at “bargain prices” from India’s most vulnerable and economically desperate people.
“The technology gains flow to wealthy US or China tech firms, while the human cost lands on the most vulnerable workers,” he said.
“The irony is brutal — workers are being paid poverty wages to train their own replacements. For India, where millions depend on manufacturing and garment jobs, this could trigger massive displacement with zero safety net.”
