The package was still in transit, but the system already knew what Chen Qing would do next.
This system was called the Predictive Engine, launched by an e-commerce platform in 2038.
The logic was simple: you ordered a bathroom shelf. The system starts reasoning — why did you buy this? Probably not enough bathroom storage. What will you need next? Towel hooks, maybe non-slip mats, maybe bathroom lighting...
While the package was still at the distribution center, the system had already prepared "things you might need next" on Chen Qing's homepage.
Chen Qing was 38, a graphic designer at a design firm, single, living alone in a one-bedroom apartment.
She never thought anything was wrong with these recommendations, until last winter.
She had ordered a book — Contemporary Chinese Picture Book Creation. A book she had browsed for a long time in a bookstore before deciding to buy.
The next morning she opened the app to check delivery status. The first thing on the homepage:
"Prepare Everything for New Life: Mother and Baby Essentials"
She froze.
Below: cribs, bottle sterilizers, postpartum meal delivery, hospital bag checklists.
She was not pregnant. She had no plans to be.
She told her colleague Lao Fang. He laughed. "The algorithm thinks you're pregnant."
"Why?"
"Picture books. Adults buying picture books — the system figures it's probably for a child, so it infers you're expecting or already pregnant, and pre-loads maternity content."
Chen Qing was silent for a long time.
Three days later the package arrived. Chen Qing tore open the envelope and took out the picture book, running her fingers across the cover.
She had bought the book because she wanted to learn to draw picture books herself. She had wanted this for years. This was the first time she had acted on it.
She opened the app, found the setting to disable personalized recommendations, and switched it off.
Then she turned her phone face-down on the desk and opened the picture book to learn its first layout structure.
The system logged in the background: user disabled personalized recommendations, behavior anomalous. Note: potential churn risk, display "enable recommendations for discount" popup on next login.
A year later, Chen Qing finished her first picture book. She self-published 50 copies and gave them to friends.
She never turned that switch back on.
It took her a long time to realize: the algorithm had not been wrong about her. It was just one step early. It knew she was about to create something — it just wasn't what the system assumed.
Before the package arrives, some things are already decided. Just not the ones the algorithm thought.
More sci-fi stories: wdsega.github.io
Top comments (0)