"Micro1 orchestrates a global effort where gig workers film themselves performing household tasks, providing essential data for training humanoid robots. This initiative highlights the economic disparity between the low wages paid to workers and the substantial investments made in robotics."
"The footage collected by Micro1 feeds into a data-hungry approach similar to that used for large language models, but focuses on real-world physical manipulation. This includes human hands performing tasks in the unpredictable environments of homes."
"Investors have poured billions into humanoid robots, while companies like Scale AI have amassed over 100,000 hours of training footage. The structural pattern shows that wealthy markets outsource the labor-intensive foundation of AI development to lower-cost regions."
A gig economy is developing globally, with workers filming domestic tasks to train humanoid robots. Micro1 recruits contractors from countries like Kenya and the Philippines to record activities such as washing dishes and folding clothes for about $15 an hour. This footage is essential for robotics companies, which require extensive real-world data for training. Despite the billions invested in robotics, the compensation for gig workers remains low compared to the value generated from their contributions, illustrating a significant economic disparity in the industry.
Read at Silicon Canals
Unable to calculate read time
Collection
[
|
...
]