Why doesn't anyone respond to your resume? You send resume after resume. The answer almost never arrives. When it arrives, it is automatic, standardized and cold: “we will not proceed with your application”. It feels like hundreds of companies have reached the same conclusion about you at the same time. A study led by researchers at Stanford University suggests a technical explanation for this experience, which is increasingly common and was already highlighted by g1 in April. Maybe you're not being rejected by several different companies, but rather the same system over and over again. ?? Do you have any reporting suggestions? Send it to g1 The research, entitled "Algorithmic Monocultures in Hiring", is the most comprehensive ever carried out on recruitment mediated by artificial intelligence. The researchers analyzed an unprecedented base of real data, with more than 3.4 million candidates and around 4 million applications evaluated in 156 companies from 11 sectors of the economy. The volume of data already draws attention, but there is an even more relevant detail: all of these applications were evaluated by algorithms developed by the same technology supplier. This made it possible to observe a phenomenon that usually goes unnoticed by candidates, companies and even job market researchers. When many organizations use similar systems to select professionals, decisions are no longer completely independent. ? The authors call this phenomenon “algorithmic monoculture”. The concept was borrowed from agriculture, in which large areas are occupied by a single crop species. Although this model can bring efficiency gains, it also creates vulnerabilities, as any problem tends to spread quickly. A Stanford study suggests that candidates may be repeatedly being rejected by the same algorithmic logic, even when applying to different companies. Pexels Different companies decide in similar ways In the recruitment and selection market, what is being standardized is not the production, but the criteria used to decide who advances or not in a selection process. For decades, hiring decisions were in the hands of recruiters, managers and teams with their own visions. Even when faced with similar CVs, it was common for them to reach different conclusions. With the expansion of automated systems, part of this diversity tends to disappear. Different companies may end up using models that analyze candidates in very similar ways, reproducing the same patterns on a large scale. In practice, job seekers may come across several seemingly independent entry points, but opened or closed by the same logic. ? This possibility led researchers to investigate a phenomenon called "systemic rejection". The term describes situations in which a candidate applies to several vacancies and is rejected from all of them. This type of experience has always existed, but what caught researchers' attention was the frequency with which this occurs when selection processes are influenced by the same systems. Data shows that around 10% of candidates who apply for four vacancies are rejected from all of them. The pattern remains even when the number of applications increases. Among candidates who apply for 10 vacancies, approximately 4% accumulate 10 consecutive rejections. At first glance, the percentages may seem modest. From a statistical point of view, however, they reveal an important pattern: rejections accumulate with a greater frequency than expected in independent decisions. To check whether this behavior could be explained by chance alone, the researchers compared the results with a theoretical baseline and with evidence from previous studies on recruitment processes without algorithmic centralization. The conclusion was clear: successive rejections are not just the result of bad luck or coincidence. They reflect an evaluation logic that is repeated among different companies. This dynamic helps explain another characteristic that is increasingly common in selection processes. In most cases, algorithms do not make the final hiring decision. They act first, as an initial filter that defines which candidates advance and which are eliminated. Therefore, many professionals can be eliminated before a recruiter even analyzes their resumes. From the candidate's point of view, the experience is silent: there is no interview, no contact with the company nor, often, an explanation for the rejection. Part of the frustration of job seekers may be linked precisely to this hidden stage of the process. The CV is sent, but does not actually compete for the position. The so-called "algorithmic monoculture" causes different employers to evaluate professionals with similar criteria, reducing the diversity of decisions. Pexels Similar profiles tend to receive similar responses Researchers found evidence that candidates with similar characteristics tend to receive similar evaluations, even when applying for different companies. ? When a system considers a profile to be poorly suited, there is a significant chance that other similar systems will reach the same conclusion. And vice versa. The bottom line is that AI models share similar classification criteria. As a result, an initial assessment, which may be limited or imperfect, gains weight when reproduced in different selection processes. Given this scenario, the researchers tested a practical question: does sending more applications still increase the chances of getting a place? The answer is yes. But this gain tends to be smaller when decisions are repeated. In simulations, a candidate would need to apply for around 10 vacancies to have a high probability of receiving at least one positive recommendation in a scenario of independent decisions. When processes are influenced by centralized systems, this number rises to around 25 applications to reach a probability of 99.9%. Market concentration magnifies the effects The results of the study do not just concern algorithms. They also raise questions about the structure of the technology market applied to recruitment. Today, many companies use solutions developed by a relatively small number of vendors. Some serve organizations from different sectors and operate on a large scale. This concentration magnifies the effects of algorithmic monoculture. When a single system influences decisions in dozens or hundreds of companies, possible failures are no longer isolated cases. The same goes for limitations or biases built into models. Therefore, researchers argue that technological concentration deserves attention not only from a competitive point of view, but also due to its impacts on professional opportunities. Despite the growing influence of artificial intelligence in selection processes, the sector still operates with little transparency, according to researchers. The authors themselves highlight that large-scale independent studies are rare. The main reason is that platforms rarely make their data available for external analysis. This creates obstacles for both oversight and the advancement of knowledge. Without access to information, it becomes more difficult to identify flaws, measure biases, and understand how these systems affect different groups. The challenge is especially relevant because these decisions have a direct impact on access to employment, income and career opportunities.