Naveen Kumar straps on a GoPro camera to his forehead in the busy town of Karur, Tamil Nadu. Taking a deep breath, he reaches into a basket for a towel, unfolds it, and starts folding it with immense precision thrice-no more, no less-and finally places it neatly aside. The sequence should not take more than a minute, and if he misses one step, he goes all over again. This ritual, repeated hundreds of times, isn’t for personal practice. It’s the raw material fueling the next wave of AI robotics.
These recordings from Indian engineers form first-person training data for robots learning household chores. By capturing human dexterity, the subtle wrist twists, fabric grips, and real-time adjustments, it helps bridge the divide between computer simulations and messy real-world environments. Robots, after all, struggle with flexible objects like laundry a slight miscalculation can mean tangles. This human-in-the-loop method, as experts call it, gives the physical AI models the nuances they need for smoother, more adaptable performance.
A tweet from X.
The story originates from a Los Angeles Times report by Nilesh Christopher, published on November 2, 2025, which profiles workers at Objectways, a data-labeling firm in Karur. As Christopher describes:
“He mounts a GoPro camera to his forehead and follows a regimented list of hand movements … If it takes more than a minute or he misses any steps, he has to start over.”
The piece has since spread via outlets like Dexerto, highlighting a global push in AI training. Objectways, with around 2,000 employees, has delivered about 200 such towel-folding videos to U.S. clients, alongside annotating 15,000 clips for other tasks.
This approach marks a clever innovation in robot dexterity training. Companies like Figure AI, Encord, Micro1, and Nvidia are investing heavily, eyeing a humanoid robot market projected to hit $38 billion by 2035, according to Nvidia estimates. The GoPro footage offers diverse, real-world data, echoing GoPro’s own opt-in program launched in July 2025, where users can monetize videos for AI model training.
Yet, beneath the tech optimism lies the human cost. These engineers perform repetitive, often tedious work data labeling and scripted actions that powers billion-dollar industries. As one Objectways worker told the LA Times, the process demands exact setups, from lighting to table color, with errors leading to discarded footage. This “invisible labor,” as commentators describe it, raises ethical AI questions: Are workers fairly compensated? What about the monotony of simulating chores they might not do at home?
Privacy adds another layer. Firms like Figure AI have explored capturing footage from inside homes potentially 100,000 of them sparking concerns over consent and data security. In a world where robots could soon enter Western households, this India-based work underscores a supply chain reliant on global labor disparities.
Ultimately, this tale from Karur reminds us that machine intelligence still draws deeply from human effort. As robots get closer to folding our laundry and tidying our spaces, the engineers behind the cameras personify the mix of creativity and grind pushing ethical AI forward. To Western audiences, it’s a look at the hidden workforce shaping the tools that might one day redefine daily life, and that begs us to think about not just what robots can do, but at what human cost.
