In a small kitchen in Chennai, Indian housewife Nagireddy Sriramyachandra records herself slicing mangoes. A smartphone is strapped to her head as she goes about routine household work. The footage is not for social media. It is used to train artificial intelligence systems.

She earns 250 rupees, about $2.60, for one hour of recording. The work may look ordinary, but the data is valuable for global tech companies building robots that learn human movement.

“They pay 250 rupees an hour just for doing housework,” she said. “That is hard to ignore.”

She is part of a growing workforce in India that supplies training data for AI systems. These workers are helping machines understand real-world actions through video, motion tracking, and first-person recording.

Everyday Work Turned Into AI Training Data

AI models can already process large amounts of digital text and images. Physical movement is harder. Developers need real-life examples to teach machines how humans behave in normal environments.

This is where egocentric data comes in. Workers wear head-mounted cameras and sensors while doing daily tasks. The recordings capture how hands move, how objects are handled, and how people interact with their surroundings.

Some workers operate from home. Others work in studios or factory-like setups. Devices often give real-time feedback during recording. One worker described alerts such as “hands not detected” when movement falls outside required patterns.

The content is then uploaded through apps to data companies. These firms operate across India and abroad and supply training data to major technology clients, including large multinational corporations.

Read More: India is teaching Google How to Scale AI in education.

India’s Expanding Role in AI Data Labor

India has become a major hub for data collection and annotation work. Thousands of workers now contribute to building datasets that support robotics and AI development.

In Bengaluru, often called India’s tech capital, researchers say this type of work is likely to grow. Digital labour expert Aditi Surie from the Indian Institute for Human Settlements noted that demand for real-world training data is increasing as AI systems expand into physical tasks.

Robotics companies are also scaling rapidly. Market forecasts suggest humanoid robots could reach widespread commercial and industrial use by mid-century. That growth increases demand for human-generated training data.

India is positioning itself as a key supplier in this supply chain, handling large parts of data generation and processing work.

Read More: Emergent, India’s Vibe-Coding Startup, Crosses $100M ARR in 8 Months

Work, Wages, and Uneven Benefits

For many workers, this new form of labor offers short-term income. Payments are small but immediate. Tasks are simple but repetitive.

Sriramyachandra works from her kitchen. Others record activities in workshops or temporary setups. The work does not require formal technical training, but it does require time and consistency.

Some workers see it as flexible income. Others see it as uncertain work tied to evolving technology systems they do not control.

A 55-year-old worker, Ponni, who makes flower garlands in Bengaluru, has also taken part in similar recording tasks. A phone is placed on her head while she works on roadside stalls.

She said younger workers may face more pressure if similar roles become widespread. “The next generation may struggle if this becomes the norm,” she said.

Automation Debate and Job Security Concerns

The rise of AI training labor has also raised wider concerns about employment. Government policy discussions in India have mostly focused on how AI will affect white-collar jobs.

A report from policy think tank NITI Aayog noted that many conversations about automation focus on skilled professionals and potential job losses in formal sectors.

At the same time, it argued that far less attention is given to informal workers, who make up a large part of India’s workforce. The report called for more focus on how AI development can benefit them directly.

Experts say the current model creates a paradox. Workers are helping build systems that may eventually replace or reshape parts of human labor, while also depending on that same system for income today.

Read More: Microsoft announces $17.5B investment in India to strengthen AI efforts

A Growing Global Pattern

India is not alone in this shift. Across different regions, human workers are increasingly used to train AI systems through labeled data, video recordings, and behavioral input.

The difference is scale. India’s large informal workforce makes it a key source of affordable training data for global AI companies.

As robotics and AI systems move closer to physical tasks, demand for real-world human data is expected to grow further. That means more workers may find themselves contributing to systems that mirror their own daily actions.

The long-term impact remains uncertain. For now, the work continues one recorded motion at a time.