Rohan Paul@rohanpaul_ai · X哈佛商业评论最新文章指出，AI正从两端破坏招聘：简历更易伪造，远程面试更易实时脚本作答。旧简历信号失效，候选人可用AI快速生成关键词丰富的申请材料，而AI筛选器反而偏爱AI风格简历——引用研究显示，此类简历入围率高出23%至60%。首轮远程面试中，AI助手可实时提供答案，尤其对冲突处理、动机回答等可预测的行为问题。伤害不仅是弱候选人被误认为强（假阳性），还有非传统候选人因简历未优化而完全被忽视（假阴性）。建议用实时工作模拟替代可预测问题：面试官在回答中途改变事实、要求候选人解释权衡并保持推理连贯。实用版本：给出杂乱工作场景，要求决策，再添加意外约束或矛盾，让候选人当场修改答案。
New Harvard Business Review article.
AI is now breaking hiring at both ends， with résumés becoming easier to fake and remote interviews becoming easier to script live.
Hiring systems now reward people who can perform the hiring process， not always people who can do the work.
The old résumé signal is weakening because candidates can generate polished， keyword-heavy applications in minutes， while AI screeners may favor text that looks like AI output， with one cited study finding 23% to 60% higher shortlisting for model-like résumés.
Remote first-round interviews are also losing trust because live AI assistants can suggest answers during calls， especially for predictable behavioral questions like conflict stories， motivation answers， and rehearsed career narratives.
The damage is not only false positives， where weak candidates look strong， but false negatives， where unconventional candidates never get seen because their documents are less optimized than their thinking.
They propose replacing predictable first-round questions with live work-simulation prompts where the interviewer changes the facts mid-answer， asks the candidate to defend tradeoffs， and checks whether their reasoning stays coherent.
A practical version is： give a messy job-relevant scenario， ask for a decision， then add a surprise constraint or contradiction and make the candidate revise their answer out loud.