The reason behind large American companies re-adopting offline recruitment models is actually
"In hindsight, the performance in that interview was too perfect" - more and more HRs tell the same story, that is, job seekers used AI extensively when applying for jobs, so that the company hired an employee who did not meet the requirements and had to fire him later.
"In hindsight, the performance in that interview was too perfect" - more and more HRs tell the same story, that is, job seekers used AI extensively when applying for jobs, so that the company hired an employee who did not meet the requirements and had to fire him later. Ben Eubanks, CEO of American human resources consulting firm Lighthouse Research & Advisory, has heard from many companies and hiring managers: Candidates are eloquent during interviews, but once they are on the job, it is revealed that those abilities are not theirs at all. Eubanks points out that in some cases, AI not only fools interviewers, but even job applicants. “AI tools give job seekers an illusion: ‘Ah, I think I can do this job’. But the cruel reality is often that they can’t do it at all.” Kiss more frogs Employers have long been accustomed to candidates polishing their resumes or reciting lines, but today's concerns run much deeper. Company managers began to fear that the entire interaction process had been intervened, optimized or even falsified by software. This leaves them unsure whether they are actually evaluating a candidate or whether they are simply being shown a software-simulated "perfect stand-in." Rich Braun, a partner at the American headhunting firm Delphi-US, lamented: “Recruiting has become much less efficient. We have to kiss more frogs in order to find the real Prince Charming.” Data compiled by Gartner, a global market research institution, at the end of last year showed that about 50% of job seekers said they had used AI at some stage in their job search, up from about 40% in 2024. Among them, 32% use AI to generate written materials required for job applications (such as accompanying portfolios), and 26% use it to cope with ability assessments; 13% of respondents admitted that they use AI chat tools in real time during remote interviews. Note: The picture shows the proportion of respondents who participated in the survey and have used AI to find jobs, and the proportion of people who use AI in various job search links. Herval Freire, chief technology officer of health-tech startup Telepatia, calls this behavior the "teleprompter phenomenon." The tell-tale signs, he said, are usually that the candidate's eyes are darting across the screen and his tone sounds like he's reading aloud. Since this happened so often, Freire began asking questions about personal hobbies during interviews in an attempt to break the candidate's trance. When they stutter and can't answer (which Freire says is often the case): "I'm like, 'Man, you're not even listening.'" There are a growing number of tools claiming to help would-be engineers cheat on coding tests and technical interviews. These services, which charge from $10 to $1,800, claim to be able to process the content and audio on the user's screen in real time to solve various programming problems instantly. Reddit forums are even filled with tips for evading detection, such as "never switch tabs," and which programs work best (which posters say are pretty average). In the fiercely competitive job market, some people say that they will do anything to gain an advantage. Their logic is that AI has to be used after joining the company anyway, so why can’t it be used during interviews? Caroline Ogawa, director of market research firm Gartner, pointed out that human resources departments have noticed an increase in the proportion of "wrong hires". These types of “bad hires” are not only frustrating, but also costly to the business. According to the Society for Human Resource Management (SHRM), the typical cost to fill an open position is about $1,300—and that's not even counting onboarding costs, weeks of lost productivity, and the time management spends searching for a replacement. Mohit Bhende, co-founder of Karat, an American technical interview outsourcing service provider, believes that intrusive anti-cheating methods such as tracking eye movements and voice patterns should be avoided in online interview systems. He said: “This creates a fundamentally antagonistic atmosphere in the interview.” In this context, some giants have drawn hard red lines.
Department store Target clearly informed candidates that their interview responses must reflect “personal understanding and judgment and must not rely on the assistance of external AI tools.” Sundar Pichai, CEO of Google’s parent company Alphabet, also said last summer that Google was leaning toward adding at least one round of face-to-face interviews “to make sure the basics are solid.” Google has also told the media that currently almost all software engineers must go through at least one round of offline confrontation. L'Oréal has classified interviews as an "AI-free zone" and requires candidates to complete at least one round of offline assessment. It is worth noting that Anthropic, a leading AI company, also explicitly prohibits candidates from using AI during on-site interviews and home interviews unless explicitly authorized. Today, job seekers wading through a job market that is already bad enough must not only prove that they stand out among competitors, but also that their perfect performance was not the work of an AI. (
Department store Target clearly informed candidates that their interview responses must reflect “personal understanding and judgment and must not rely on the assistance of external AI tools.” Sundar Pichai, CEO of Google’s parent company Alphabet, also said last summer that Google was leaning toward adding at least one round of face-to-face interviews “to make sure the basics are solid.” Google has also told the media that currently almost all software engineers must go through at least one round of offline confrontation. L'Oréal has classified interviews as an "AI-free zone" and requires candidates to complete at least one round of offline assessment. It is worth noting that Anthropic, a leading AI company, also explicitly prohibits candidates from using AI during on-site interviews and home interviews unless explicitly authorized. Today, job seekers wading through a job market that is already bad enough must not only prove that they stand out among competitors, but also that their perfect performance was not the work of an AI. (