Courtesy the Instagram handle of The Indian Express, the sight of factory workers wearing head cameras to train artificial intelligence should trouble anyone who cares about dignity at work. What looks like a modern data-collection exercise is, in reality, a new form of labour extraction, where the human effort is being captured, packaged, and fed into machines that may one day – very soon – replace the very people who supplied the knowledge. In a textile factory in Nagpur, worker Ashish Narayan reportedly described the experience in a bitter phrase when he said that it feels like “working in your own grave, while you make your own casket.”
This is so powerful because it says what many people hesitate to admit openly, and that is the AI economy can depend on human skill while quietly erasing human value. The practice at the centre of this issue is known as ‘egocentric data’, which means first-person recordings of how people perform physical tasks. Cameras worn on the forehead or body capture movements, gestures, and decision-making patterns so machines can learn to imitate them. This is not just a technical process. It is also a workplace relationship. When workers do not know what is being recorded, where it goes, or how it will be used later, the line between training and surveillance becomes dangerously thin.
Advertisement
The first ethical issue is ‘consent’. If workers fear losing their jobs or already earn very little, they cannot truly refuse. So “voluntary” participation can be coercive in practice. The second issue is ‘ownership’. When a worker’s movements, skills, and speed train a profitable AI, who gets the value? If only the company benefits, the worker has effectively given unpaid intellectual labour in physical form, while receiving only a small hourly wage and no share of long-term gains. The third issue is ‘privacy’. Constant recording in factories can capture more than work tasks, revealing habits, relationships, stress, and sensitive personal details about workers and their safety, per se.
Advertisement
This raises a deeper concern for India, which has become a battleground for global AI firms looking to enter enterprise markets, win developers, and lock businesses into their ecosystems. That competition may create jobs in the short term, but it also risks normalising a model where workers are turned into data sources before being replaced by the systems they helped create. The danger is not limited to factories. If this logic spreads, it could reshape how we think about labour itself. Human effort may no longer be valued for the dignity it carries, but only for the data it can generate.
That is a dangerous shift in a country where inequality already affects who gets secure work, who gets lower wages, and who gets listened to. This practice bypasses basic intellectual property fairness. Decades of specialized, tactile experience, such as the nuanced, unwritten intuition a human worker develops over a lifetime of labour, are extracted without fair compensation or long-term residuals. Critics argue that this process functions as a systematic expropriation of human craftsmanship, commodifying highly precise labour to feed growing corporate capital.
The trajectory of human progress has always been mirrored by the tools we create. From the steam engines of the Industrial Revolution to the silicon chips of the digital era, technology has consistently redefined the boundaries of work, society, and identity. Today, we stand on the precipice of a far more profound shift, encircled by the ever-expanding rise of Artificial Intelligence (AI) and advanced robotics. No longer confined to executing pre-written code or mathematical scripts, modern AI models can reason, generate creative content, and systematically learn to replicate complex physical tasks. Yet, as AI deepens its roots into our daily workflows, it brings along an uncomfortable dynamic.
The rapid deployment of these technologies has outpaced our legal framework and ethical vocabulary. We find ourselves balancing between an automated future of unprecedented productivity and a dystopian landscape characterized by labour exploitation, the erasure of individual agency, and a fundamental misunderstanding of what makes human expression unique. Beyond the immediate social and economic strains, the unchecked expansion of artificial intelligence faces a more volatile, physical constraint – that of geopolitics and energy. The narrative surrounding AI often paints it as an ethereal, weightless cloud of pure intellect.
In reality, it is an immensely fragile physical infrastructure business that relies heavily on cheap, stable natural resources and highly concentrated supply chains. Recent global instability has exposed this reality. The US push on technology relies on huge, long-term investments, amounting to many billions of dollars from Middle Eastern state funds. But local political conflicts in that region have put those investments at risk. At the same time, data centres are hitting big energy shortages at home. In the US, about 40 per cent of server power comes from natural gas. As geopolitical tensions raise fuel prices, the cost of running these centres have gone up sharply.
Beyond the hype, more than 80 per cent of corporate AI projects collapse before or immediately after the proof-of-concept stage because of poor data integration and huge upfront costs. Crucially, when algorithms leave structured digital sandboxes, such as code repositories, curated text databases, or predictable logistics grids and confront the chaotic, messy reality of physical, emotional human life, they fail spectacularly. A viral video from a “Future Era” robot store in Shenzhen captures this limitation. A multimillion-dollar humanoid robot took the stage to perform Michael Jackson’s 1983 hit “Billie Jean.” At first it managed a few preprogrammed, passable moves. But when it stepped onto a slightly raised edge of the stage, its balance system collapsed.
The robot stumbled in jerky, mechanical contortions. The act ended not with applause but with a technician walking onstage, seizing the inert machine by its frame, and dragging it off like broken office equipment. It is implied that genuine artistic expression depends on subtle micro-dynamics – facial shifts, the intentional catching of breath, and the charged weight of silence. Those cues arise from an interior emotional life like grief, joy, defiance or v ulnerability. L acking inner consciousness, machines can only mimic such forms, producing an uncanny, hollow caricature or mimicry. Artificial intelligence is neither an omnipotent digital messiah nor a harmless, cute novelty designed for social media entertainment.
It is a highly powerful, deeply flawed, and resource-intensive automation tool. If we are to build an ethical future, our regulatory priorities must shift away from abstract sci-fi worries of machine sentience and focus directly on tangible human protections. We must establish rigid legal firewalls around data collection, ensuring that egocentric data harvesting can never be used as a coercive tool to strip workers of their livelihood without clear compensation and long-term societal support. We must look at the technology for what it truly is, that is, learning algorithms and machine inferences that excel at sorting massive datasets and automating repetitive digital tasks, but remain completely blind to the deeper textures of human experience. By maintaining a sharp skepticism toward its limits and an unyielding commitment to human rights, we can prevent AI from becoming a tool of absolute exploitation, ensuring instead that it remains a well-regulated assistant to human ingenuity.
(The writer is Programme Executive, Gandhi Smriti and Darshan Samiti.)
Advertisement
