In an intriguing twist on the gig economy, certain AI startups are now incentivizing individuals to record themselves engaging in everyday tasks like folding laundry and loading dishwashers. Companies such as Encord and Micro1 are catering to the burgeoning demand from robotics firms seeking high-quality training data necessary for advancing their technologies. Robotics is rapidly emerging as a key sector within the AI landscape, with notable investors, including Vinod Khosla, forecasting a transformative period akin to the rise of ChatGPT. This year alone, venture capital funding in robotics has soared to $12.1 billion, according to PitchBook data, signaling a robust interest in the field. However, the robotics sector faces a significant challenge: generating the appropriate training data. Unlike large language models that can utilize vast amounts of internet data, robots require specific real-world data to master fundamental skills. "Unlike LLMs, robotics doesn't have the internet as a ready-made dataset — you have to generate training data from scratch in the real world, which is far harder," stated Ulrik Hansen, co-founder of Encord. He noted that demand for such data has quadrupled compared to the previous year. For those willing to participate, the compensation is quite appealing. Individuals can earn between $25 and $50 per hour for tasks like folding clothes while donning Meta's Ray-Ban glasses to capture the activities on video. In more specialized scenarios, such as demonstrating the use of surgical equipment, pay can rise to as much as $150 per hour, according to Hansen. Other players in this space, like Scale AI, are also stepping up efforts in the robotics domain, having launched a dedicated lab in San Francisco that has already generated over 100,000 hours of training footage. The demand for this data is evident from collaborations with companies like Physical Intelligence, a humanoid robotics firm that counts Amazon founder Jeff Bezos among its backers, and Boston Dynamics, famous for its robotic dogs. Despite the progress, acquiring sufficient training data remains a daunting task. One startup even turned to Craigslist, offering $10 to $20 per hour for individuals to film themselves completing chores on their iPhones. The founder of this stealth-mode company expressed the ongoing struggle to secure quality training data, highlighting that existing datasets are often insufficient, averaging around just 5,000 hours. As the robotics field continues to evolve, the innovative approaches taken by these startups may well pave the way for significant advancements in automation and AI technology.
Tesla has experienced a significant boost in its electric vehicle sales in China during the initial two months of 2026, ...
CNBC | Mar 13, 2026, 07:20
The International Imaging Technology Council (Int’l ITC) has raised concerns against HP regarding recent firmware update...
Ars Technica | Mar 12, 2026, 20:35
Google Maps is set to revolutionize the way users navigate their surroundings with the introduction of innovative AI-dri...
Business Today | Mar 13, 2026, 06:00
Truecaller, the popular caller identification platform, has introduced an innovative feature designed to help families p...
TechCrunch | Mar 13, 2026, 04:45
In a significant shift for the company, Adobe has announced that its long-serving CEO, Shantanu Narayen, will be steppin...
Business Today | Mar 13, 2026, 03:15