The System Wasn't Fair Before AI Either
On algorithms, rejection, and what recruitment has never been honest about.
One of the most common conversations I have with clients goes something like this:
“I know I can do the job. So why am I not getting interviews?”
For years, the assumption was that someone in HR or a hiring manager had looked at the application and decided it wasn’t the right fit. That was hard enough to hear. But increasingly, it isn’t even what’s happening.
In many organisations today, your application is being assessed by software before a recruiter or hiring manager ever sees it. Sometimes that software is simply organising applications. Sometimes it is ranking them. Sometimes it is deciding which candidates progress and which do not, with no human involved at any point.
And here is the uncomfortable truth that sits underneath all of it: a rejection is not always a judgement on your capability. Sometimes it is a judgement made by an imperfect system.
What the Research Actually Shows
In May 2026, Stanford University published one of the largest independent studies of AI hiring systems conducted to date. The research examined more than four million job applications across 156 employers.
The findings were significant. More than 90 per cent of employers were using some form of automated screening before human review. Many were relying on the same small group of technology providers — which means candidates may be assessed repeatedly by similar systems across multiple organisations, without ever knowing it.
Evidence of discriminatory outcomes was identified across a number of job categories. But perhaps the most striking finding was this: every applicant in the study was recommended by at least one AI model.
Every applicant. Not a select few. Not the strongest candidates. Everyone.
The same candidate could be rejected by one system and recommended by another. Different systems reached different conclusions about the same people.
As somebody who has spent more than twenty years assessing talent, I find that extraordinary. Not because it proves AI is broken but because of what it reveals about the nature of assessment itself.
What Is Happening in Europe
Research on this side of the Atlantic is reaching similar conclusions.
The FINDHR consortium, which includes Radboud University in the Netherlands, found significant differences in outcomes when identical CVs were submitted with different names. Nothing changed apart from the perceived background of the applicant. The BIAS project reached a comparable conclusion: bias does not simply appear after a system is deployed. It can already be embedded in the data used to train these systems in the first place.
If the underlying data reflects historical patterns or societal stereotypes, technology does not remove those patterns. It reinforces them.
The Netherlands has already seen how damaging automated decision-making at scale can become. The childcare benefits scandal demonstrated what happens when flawed assumptions are baked into systems that affect people’s lives. That was not a recruitment example, but the lesson transfers.
The European Union has responded with regulation. From August 2026, AI systems used in recruitment and employment decisions fall within the EU AI Act’s high-risk category. For employers, this creates additional obligations around testing, transparency, and oversight. For job seekers, it creates greater protection including stronger rights to understand when AI is being used as part of an assessment process and to challenge decisions where appropriate. Human oversight becomes a requirement rather than a choice.
But Here Is What I Actually Think
Spend enough time on LinkedIn and you could be forgiven for thinking AI has suddenly made recruitment unfair.
I am not convinced.
What I think AI has done is expose something that many recruiters, hiring managers, and candidates have known for years but rarely talk about openly.
Hiring has never been as objective as people would like to believe.
For years, candidates have been told that if they were rejected, someone else was simply a better fit. Sometimes that is true. Often it is not.
Over the course of my career, I have seen excellent candidates rejected because the budget disappeared halfway through a process. I have seen roles cancelled after weeks of interviews. I have seen internal candidates emerge at the eleventh hour. I have seen hiring managers change their minds. I have seen businesses alter direction. I have seen politics influence decisions. I have seen timing work in one candidate’s favour and against another equally capable person.
None of this is new.
What is new is that some of these decisions are now being made — or influenced — by technology.
The Stanford research makes that visible in a way that human decision-making never quite did. When an algorithm rejects someone and another recommends them, the inconsistency is documented. When a hiring manager does the same thing, it tends to disappear quietly into a “thanks for your time” email.
CVs are flawed. Interviews are flawed. Psychometric assessments are flawed. Human judgement is flawed.
I have said for years that whoever cracks online recruitment will crack online dating and vice versa. The problem is structurally identical: you are trying to predict human compatibility from a profile, a few data points, and an algorithm that has never met either party. We are now seeing both industries arrive at the same uncomfortable conclusion at roughly the same time. The promise was objectivity at scale. What we got was bias at scale, dressed up in the language of optimisation.
The question is not whether AI makes mistakes. The question is whether we expect technology to achieve a level of consistency that recruitment itself has never achieved and whether we are prepared to hold it to account in a way we have never held human decision-making to account.
That is where I think many of the current debates miss the point.
What This Means for You
The danger, as I see it, is that candidates start treating algorithmic rejection as a verdict on their ability.
It is not. It never has been.
Understanding how the system works gives you an advantage. Knowing that ATS software exists, that keyword optimisation matters, that certain formats screen better than others; all of that is worth knowing. But none of it changes the fundamentals.
Build skills. Develop relationships. Communicate your value clearly. Put yourself forward for opportunities even when you are not certain you will get through. Rejection whether it comes from a human or an algorithm has never been a reliable measure of potential.
Careers are rarely defined by a single rejection. They are shaped by what happens next.
References
1. AI Hiring Tools Can Yield Racial Bias and Systemic Rejection - The first large-scale study of hiring algorithms in the wild finds concerning patterns to how system...
2. [Revisión de artículo] Algorithmic Monocultures in Hiring - This paper investigates the phenomenon of algorithmic monoculture in hiring, hypothesizing that when...
3. Q&A | Algorithmic Monoculture in Hiring - The World Economic Forum estimates that more than 90% of employers use automated systems to screen j...
4. Stanford Digital Economy Lab's Post - ✍ Algorithmic Monoculture: New study shows the impact of hiring algorithms on job opportunities Over...
5. [2605.27371] Algorithmic Monocultures in Hiring - Many employers screen job applicants with algorithms built by the same few algorithm vendors. We hyp...
6. Project Overview - BIAS - A four-year project, funded by the European Union's Horizon Europe Research and Innovation program t...
7. New Horizon Europe project looks into discrimination in algorithmic ... - FINDHR, a research project aimed at preventing, detecting and mitigating discrimination in algorithm...
8. BIAS IN ALGORITHMS ARTIFICIAL INTELLIGENCE AND ... - This EU Agency for Fundamental Rights (FRA) study examines how algorithmic bias arises and when it t...
9. AI Hiring Program (EU) - impactmania - We conducted qualitative and quantitative experiments to test algorithmic hiring tools such as Appli...
10. EU Fundamental Rights Agency report on AI algorithm biases - MIAI - The FRA's report was motivated by the need to investigate how AI can lead to fundamental rights viol...
11. EU AI Act & Recruitment: ATS Compliance Checklist Before ... - EU AI Act compliance for HR: high-risk AI classification, ATS audit checklist, data transparency obl...
12. Article 6: Classification Rules for High-Risk AI Systems