According to According to the According to the World Economic Forum 2025, more than 90% of employers already use automated systems to filter responses, and 88% of companies use AI at the initial screening stage. Machines search for keywords, check the structure and filter out anything that does not meet the specified parameters.
As investing.com notes, the reason is simple – scale. Companies physically cannot cope with the flow of candidates. For example, Goldman Sachs received received more than 315 thousand applications for internships. Google annually receives about 3.3 million responses, and McKinsey & Company – over 1 million.
Even if a recruiter wants to, he cannot study such a flow. After all, when a popular job posting gets 300-500 responses in the first three days, a recruiter has 30 seconds to two minutes to review each resume. Simon Taylor, a former recruiter at Disney, says that the figure is more like three to five seconds.
“The black hole of responses
AI was expected to speed up and simplify hiring. In practice, candidates are increasingly facing the opposite effect: responses go into a “black hole”, and strong specialists do not pass the filter due to a lack of the right wording.
The situation is supported by data: according to a Zety report published via CNBC in August 2025, one in five laid-off Americans sent more than 100 applications before finding a job. Adecco CEO Dennis Mashuel estimates the average is even higher – around 200 responses per offer.
Candidates are increasingly receiving automatic rejections literally minutes after sending a resume – too fast for a human to review.
When AI helps and when it hinders
That said, the benefits of technology can’t be completely denied. An experiment by researchers from the MIT Sloan School of Management on the basis of a large freelance platform showed that algorithmic prompts in resume writing reduce the number of errors by 5%, increase the probability of hiring by 8%, and increase the salary by about 10%.
Candidates themselves are also actively using AI. For example, 68% of job seekers are already using it for resume writing. About 67% use it to write cover letters. And 90% of managers consider it acceptable.
In fact, the market has arrived at a new norm: AI has found itself on both sides of the process.
Why resumes lose meaning
But this “algorithm race” has side effects.
An analysis of data from Freelancer.com (61,000 jobs and 2.7 million applications for 2021-2024) shows that personalized cover letters have become practically worthless. What used to be an advantage now doesn’t work, because such texts are massively generated by neural networks.
Moreover, the very balance of the market has changed: the top 20% of candidates are hired 19% less often. And the worst 20% are hired 14% more often
The most unexpected effect is the bias of the models themselves.
A study published by the Association for Computing Machinery (ACM) in 2025 found: AI systems are more likely to favor resumes written by other AIs. Models are able to recognize “their” style and systematically select it.
In addition, algorithms inherit human biases. A famous example is Amazon’s matching system, which favored men because of training sampling.
Against this background, experts advise creating AI agents that will be able not just to screen candidates, but to conduct a dialog: explain requirements, answer questions and give feedback.
