Housework Pays: Embodied AI Brings Data Collection Jobs to Communities
During a recent visit to Taishan District, Tai'an City, Shandong Province, a special community class was underway. The data collector training class, organized by Shandong Shucaigu and the local human resources and social security department, was attended by ordinary residents from the community. From device operation to details to note during collection, participants listened carefully. Since June, such training classes have entered more than a dozen communities and held many rounds.
It was through this training that 46-year-old Chen Junxi became a home-based data collector for Shandong Shucaigu a month ago. Every morning, he puts on a collection device and busies himself with housework: washing vegetables, cooking, and tidying the house. These daily life scenes are recorded by the device. The footage is submitted to Shucaigu, processed and annotated, and turned into video materials for training robots.
Chen's circumstances matched the job well. In the past two years, he has needed to care for his elderly mother and could not go out to work. With the ability to work from home and a decent income, he recommended the job to several relatives. Like him, more than 200 people in Taishan District are now engaged in this data collection. The job requires no leaving home, no fixed working hours, and no professional background. For people who need to care for family members or cannot work full-time outside, the threshold is relatively low. Collection quality is ensured through training: what scenes to collect, how to operate the equipment, and what details to note all have standards.
Why choose a resettlement community? Shandong Shucaigu's main business is providing training data to AI companies. In its early planning, it decided to collect data in real environments as much as possible. After multiple surveys, it discovered Bocheng Community in Tai'an. The community resettles residents from eight surrounding administrative villages, with more than 5,000 households and a population of over 18,000. It is the largest resettlement community in Taishan District. The abundant human resources and diverse real collection environments made it convenient for the company to settle in. In April this year, Shandong Shucaigu chose to locate in Bocheng Community. The human resources department publishes information and recommends jobs through employment service stations, and organizes interested people across the district for on-site visits and skills training.
This siting logic corresponds to data collection's demand for real environments. Laboratories and specially built training grounds can collect data under controlled conditions, but the clutter, variability, and unpredictability of family environments cannot be reproduced in a lab. In resettlement communities, different families' living habits, housing layouts, and object placements vary. These differences themselves are variables that training data needs to cover.
The scenes are expanding from home to supermarkets, pharmacies, and villages. Qiu Wenjing, 32, is a mother who has been a home collector for nearly two months. Recently she added a scene collecting in a supermarket. Every day after sending her child to school, she takes a collection device and goes with several neighbors to a supermarket to arrange fruit and place goods, recording shelf-stocking videos. The local government helped the company coordinate with supermarkets and pharmacies, and video collection has extended from home environments to surrounding businesses. Qiujiadian Town has also begun expanding collection scenes to villages. Wanglinpo Village is a traditional village that has developed rural tourism in recent years, building workshops for making tofu, steaming buns, and baking pancakes. Craftspeople wear collection devices while working, recording operation processes while making traditional foods.
Starting next month, Shandong Shucaigu will add 2,000 devices in succession. This means 2,000 people will be able to find new digital jobs at their doorsteps. Whether this model can continue depends on whether embodied AI companies' demand for real-scene data continues to grow. If robot training demand for home scenes, retail scenes, and traditional handwork scenes continues to expand, the community data collector job will continue to exist.