Data annotation, the work of processing texts, photos and videos for artificial intelligence (AI) recognition, is not a sophisticated task. But it requires legions of cheap graduate level roles in China to help machines and robots understand and interpret data. For instance, workers at smart technology companies are employed to help AI memorize human faces by marking key facial characters. While working to improve AI, workers are asked to put themselves into the AIs’ collective mind even though human minds are individual. Different from robots, humans can command tacit knowledge when learning from their personal experiences when even comprehension defies description. Managers at AI companies are trying to keep an equilibrium between the need to improve AI and job satisfaction for their employees. The staff face high turnover, as these roles contributing to the digital world may prove dull and valueless to personal growth.