AI Hiring Tools Meet AI-Wielding Job Seekers in China
Job hunting in China has become a contest between two kinds of software. Companies are deploying AI to sort résumés, run first-round interviews and score candidates on speech and facial cues. Applicants, in turn, are turning to AI tools that rehearse likely questions, suggest answers and even generate polished self-introductions. What was once a conversation between people is increasingly a negotiation between models.
The shift is partly a matter of scale. Large employers can receive thousands of applications for a single opening, and automated screening promises speed and consistency. Recruiters say AI interviews reduce scheduling costs and filter out obviously unsuitable candidates before a human ever joins the call. For fresh graduates applying to dozens of companies in a single season, the efficiency cuts both ways: the same technology that speeds up their applications also speeds up their rejections.
Candidates have adapted quickly. Online forums and short-video platforms are full of advice on how to beat an AI interviewer: look directly at the camera, keep answers to a tight structure, avoid long pauses, and practice with a chatbot that mimics the company’s question bank. Some job seekers describe using AI to draft answers to common behavioral questions, then memorizing the phrasing. The tools are cheap and widely available, which makes the countermeasure as accessible as the screening.
The result is a strange kind of symmetry. An AI interviewer assesses a candidate who has been coached by AI, and both sides are optimizing for signals that neither fully trusts. Recruiters complain that rehearsed answers reveal little about a person’s real abilities. Applicants complain that algorithmic judgments are opaque, that a bad camera angle or an unusual accent can count against them, and that no one explains why they were rejected. The frustration is less about being evaluated by machines than about being evaluated by machines nobody can question.
For employers, the appeal is not only cost. Standardized interviews produce comparable data across candidates, and in a market where hiring volumes swing sharply, that consistency has value. But the spread of AI screening is also drawing attention to what gets lost: the unscripted exchange in which an interviewer notices that a candidate thinks differently, or a candidate realizes the job is not what the posting promised. Those moments are hard to automate, and harder to measure.
Some companies are responding by mixing methods. A growing number use AI only for the earliest filter, then hand shortlisted candidates to human managers for longer conversations. Others are experimenting with structured tests that are harder to game than a spoken answer. The common thread is a recognition that the tool is a gatekeeper, not a judge, and that the final decision still needs a person willing to take responsibility for it.
The broader lesson may be about trust in automated decisions. As AI moves from résumés into live interviews, the question is not whether the technology works, but whether the people on both sides of it believe the outcome is fair. Candidates who feel reduced to a score are unlikely to accept a job offer with enthusiasm, and companies that rely too heavily on screening may find they have optimized for people who are good at being screened. The arms race is likely to continue, but its costs are beginning to show.