Can AI Really Predict Employee Work Status? Inside Baidu’s New Patent

Recently, a series of suicide resignations at PDD has brought the controversial 996 work culture back into focus.

Programmers, as the main force behind 996, have experienced overtime and shifting client demands firsthand.

Amidst this, a question arises: can we accurately reveal each employee’s work status? Baidu has recently filed a patent that claims to predict employee work status.

The patent applicant is Baidu Online Network Technology (Beijing) Co., Ltd., titled "Method, Device, Electronic Equipment, and Storage Medium for Predicting Employee Work Status," published on January 12, 2024 (CN112215447A).

It obtains the employee’s organization information, position level, and communication data; extracts a matching representation vector between employee and organization; generates temporal features; and predicts the employee’s work status.

This description sounds abstract, but essentially it uses big‑data‑driven quantitative analysis to extract person‑organization matching representations and their impact, enabling efficient, accurate, real‑time extraction of organizational fit and dynamic analysis and prediction of employee behavior.

In plain terms, it monitors work behavior and content, potentially comparing it with job level, raising concerns about surveillance and exploitation of employees’ time.

Baidu’s official statement on January 13 clarified that the patent is a “person‑job matching” management tool aimed at attracting, cultivating, and retaining talent, and has nothing to do with the 996 issue.

The debate continues: is this technology meant for supervision or other purposes?