Manman studied graphic design in college and had been working in design-related roles after graduation. She quit her job to take care of her child, but when she tried to return to the workplace, she found everything had changed completely.
"AIGC has become increasingly mature, and with my several-year career gap, it's extremely difficult for me to find a design job now," Manman explained.
After several twists and turns, she finally found a job as an AI Trainer at a major Chinese tech company. "There are two reasons why I chose this job: first, this is not an outsourcing role, I am directly employed by the first-party company; second, this job rarely requires overtime, so I have time to take care of my child."
Her work involves the AI labeling process: the upstream party is the demand side — the product and R&D teams, who put forward demands based on problems fed back from product usage. Manman's role is in the middle, where the core task is to evaluate and train the model according to demands from the product side, and formulate rules for data processing. Downstream are AI Labelers, who perform more hands-on labeling work following the established rules.
To make the transition smoothly, Manman paid out of her own pocket to take specialized courses at a training institution. "The training courses taught me basic concepts of data labeling, workflows, data processing knowledge, and so on. Personally, I think it was very helpful for my career transition, because my educational background had absolutely nothing to do with these fields."
Talking about her current job, Manman can't say she is particularly fond of it. "My previous jobs were all divergent, relying on creativity and ideas. But this job is completely different. It makes me feel like a line worker on a production assembly line. Most of my work is coordination: aligning with upstream teams on demands, and aligning with downstream teams on workflows. I only need to understand, convey, and follow the rules."
She has a very motivated colleague who also has a liberal arts background and transitioned upstream to become a product manager two years ago. But Manman herself isn't sure about that path: "I really can't say I love this job — I just want stability."
Talking about people without a science or engineering background transitioning to AI jobs, Manman reflects: "It's less about me taking the initiative to transition, and more about being pushed forward by the times. Regarding data labeling, I heard a joke a long time ago — artificial intelligence means that the amount of 'artificial' (human labor) determines the level of 'intelligence.' After actually taking this job, I deeply understand this saying."
"If you want a stable position, you definitely need to make choices that align with the general trend of the times. But if I could follow my own heart completely, I would still prefer the era before AI existed."