AI Training in Indian Factories Sparks Workforce Concerns

In Karur, India, workers like 40-year-old Shoba Kirubakaran pack textiles while wearing GoPros, capturing their actions to train AI robots. This technology, used in a Texworth factory, records workers’ movements without explicit knowledge of its purpose. Kirubakaran, earning $90 a month, discovered from interviews that her job practices help develop AI capabilities.

The emergence of videos showing Indian workers using headsets for these recordings has raised public concern. These videos sparked discussions about how AI technology influences employment in India, which has a workforce of 600 million people. Prathosh A.P., from the Indian Institute of Science, explains that AI development relies on ‘egocentric data,’ footage from the perspective of human workers. This data trains AI to replicate intricate human motor skills.

Some workers willingly record themselves for additional pay, but many, like Kirubakaran, do so without a clear understanding of the project. Astha Kapoor from the Aapti Institute describes this process as labor exploitation. The Indian workforce faces challenges such as job retraining and adapting to technological shifts, notes economist Amit Basole from Azim Premji University.

Challenges in AI Robot Deployment

While AI robots promise future efficiencies, experts like Kapoor argue that high costs and power demands limit immediate mass deployment in India. Factories like Objectways collect egocentric data for AI development but require continuous power, which remains costly and inconsistent.

Ravi Shankar, CEO of Objectways, believes that AI robots are currently more expensive compared to human labor. His company focuses on designing robots for niche tasks—such as tapping rubber trees or cleaning solar panels in remote areas—that are hard to fill with human workers. Rajeshkumar Ramaswamy of Objectways highlights how workers simulate jobs using robotic tools to collect ‘exocentric data,’ training AI to recognize and manipulate objects like water bottles.

Impact on Domestic Tasks

In training AI for household tasks, Objectways employees record themselves performing everyday chores. This data aids in developing robots for domestic use, with recordings tailored for both Indian and American preferences due to differences in household items.

Ramaswamy explains that teaching robots involves varying degrees of grip based on material, preventing damage. Employees like Sneha Jegadish participate in these recordings, earning $2.60 an hour, as they perform chores unfamiliar to them before joining the project.

Future of Workforce and AI

For workers like Kannagi Periyasaamy, paid for documenting their tasks, the reality is starting to sink in that AI may soon replace manual labor. She plans to shift to farming if her current job becomes obsolete. Yet, Periyasaamy takes pride in contributing to the training of AI, seeing the robots as students she can guide.