Your Resume Was Rejected in Seconds by a Machine That Can Not Explain Why
Your Resume Was Rejected in Seconds by a Machine That Can Not Explain Why
Part of “The AI You Don’t See” series by Akshay A. Walimbe
You spent four years getting an engineering degree. You spent two more building skills on the job. You spent three hours writing your resume, checking every line, getting the formatting right, choosing the right action verbs because someone on LinkedIn told you that “spearheaded” is better than “led.”
You applied to forty seven jobs last month.
Thirty nine of them rejected you before any human being read a single word you wrote. The rejection happened in seconds. Not because a recruiter glanced at your resume and decided you were not a fit. Because an algorithm parsed your document, scored it against a set of criteria you will never see, and sorted you into the reject pile before your application finished uploading.
You were not reviewed. You were processed.
The Machine Between You and Your Next Job
Most job seekers have heard of Applicant Tracking Systems ATS software that companies use to manage the flood of applications. What most people do not realise is how much these systems have changed.
The old ATS was basically a filing cabinet. It stored resumes, let recruiters search by keyword, maybe sorted by date. A human still made the decisions.
The new ATS is something else entirely. It is an AI powered screening system that reads your resume, analyses it against the job description, scores you on a scale, and makes a recommendation: advance or reject. Some systems go further. They assess your personality from video interviews. They evaluate your “cultural fit” based on linguistic patterns. They predict how long you are likely to stay at the company based on your job history.
And here are the numbers that should trouble you. According to a 2025 ResumeBuilder survey, 83 per cent of employers are projected to use AI to screen resumes, even though 67 per cent of them acknowledge bias concerns. Not just use it. Rely on it. Twenty one per cent of companies automatically reject candidates at every hiring stage without any human review at all. Half of all companies use AI exclusively for rejections during initial screening.
That means if you applied to a hundred jobs, fifty of them may have rejected you without a human being ever knowing you existed.
What the Machine Sees (and What It Doesn’t)
The system parses your resume, extracts text, and maps your information into fields: education, work experience, skills. If your formatting is unusual a creative layout, a two column design the parser may scramble the data. Your carefully crafted resume might become garbage text before any analysis begins.
Then comes the scoring. The AI compares your parsed resume to the job description, looking for keyword matches, semantic similarity, and years of experience. Some systems use large language models the same type of AI that powers ChatGPT which means they absorb the biases embedded in their training data.
A study by the University of Washington, presented at the AAAI/ACM Conference on AI, Ethics, and Society in October 2024, tested three state of the art large language models used for resume screening across 120 first names and more than 500 real job listings over three million comparisons in total. AI screening tools favoured white associated names 85 per cent of the time. Male associated names 52 per cent of the time. Black male candidates were never favoured over white male candidates in any test. Racial parity occurred in only 6.3 per cent of tests.
The machine that decides whether your resume gets seen by a human is, in a peer reviewed study, biased by the name at the top of the page.
The Indian Job Market and the Invisible Filter
Now let me bring this home to India.
India produces millions of graduates every year. The competition is brutal thousands of applications for the most desirable positions. Companies adopted AI screening out of necessity. When you receive ten thousand applications for twenty positions, you need a filter. But what does that filter actually filter?
Consider what an AI screening tool trained on historical hiring data from an Indian IT company would learn. It would learn that most successful hires came from a handful of premier engineering colleges. That candidates from Bangalore, Hyderabad, Pune, and Mumbai had higher retention rates. That certain English vocabulary patterns correlated with getting hired.
What is it actually learning? Not merit. It is learning existing patterns of privilege and replicating them, at scale, faster than any human recruiter ever could.
If you went to a state university in Jharkhand instead of IIT Bombay, the algorithm has already downgraded you before it reads your skills section. If your resume was written in the English of your local coaching centre rather than the English of a Bandra based resume consultant, the semantic matching may score you lower.
None of this is deliberate malice. Workday’s own Chief Responsibility Officer has said that “Workday AI does not make hiring decisions and is not designed to automatically reject candidates,” and that customers maintain human oversight. Other vendors like Amazon, after scrapping a biased hiring tool in 2018, rebuilt their systems with explicit fairness testing. The intentions are often genuine. But the system discriminates anyway, because it learned from a world that already discriminated. The AI does not create the bias. It industrialises it.
The Explainability Gap
Here is the part that should make you angry.
When a human recruiter rejects your resume, they can, at least in theory, explain why. “We needed someone with more experience in cloud infrastructure.” “Your background is more research oriented than what we’re looking for.” You might disagree with the reason, but at least there is a reason. A human judgment you can understand, challenge, or learn from.
When an AI screening tool rejects your resume, what explanation do you get?
Nothing. A form email. “After careful consideration, we have decided to move forward with other candidates.” That is it. No score. No criteria. No explanation of which keywords you were missing, which patterns did not match, which invisible threshold you fell below.
The Center for Democracy and Technology reviewed HireVue’s AI explainability statement HireVue being one of the largest AI hiring platforms in the world and concluded that it “mostly fails to explain what it does.” This is not an obscure startup. This is a company whose AI has been used to interview and assess millions of candidates. And even their official explanation of their own system was found to be inadequate.
In 2025, the ACLU of Colorado filed a complaint alleging HireVue’s platform discriminated against deaf and non white individuals. The complaint described an Indigenous and deaf applicant who was rejected and given AI-generated feedback to “practice active listening.”
Practice active listening. From a machine assessing a deaf person. Nobody reviewed this before it was sent. Nobody caught the absurdity. Because nobody was looking. The machine handled it.
The Lawsuit That Changed the Rules
There is one case that every job applicant should know about. In February 2023, Derek Mobley filed a lawsuit against Workday the enterprise software company whose platform handles hiring for thousands of major employers. Mobley alleged that Workday’s AI-based screening tools discriminated against applicants based on race, age, and disability.
In July 2024, Federal Judge Rita Lin made a ruling that had never been made before. She determined that Workday’s software was “not simply implementing in a rote way the criteria that employers set forth, but is instead participating in the decision making process.” This meant Workday itself the AI vendor, not the employer could be held liable for discrimination. It was the first time a federal court applied agency theory to hold an AI vendor directly responsible for discriminatory hiring decisions. The EEOC filed an amicus brief in support of the plaintiff’s position.
In May 2025, the case was certified as a nationwide collective action covering all applicants aged 40 and older who were denied employment recommendations through Workday’s platform since September 2020. Workday represented in court that 1.1 billion applications had been rejected using its software during the relevant period. Billion, with a B.
A billion rejections. And only now, after a lawsuit, is anyone asking whether those rejections were fair.
The Question Nobody Answers
In India, we do not yet have a Mobley v. Workday. We do not have an Illinois AI Video Interview Act requiring employers to disclose when AI is making hiring decisions. We do not have a New York City Local Law 144 requiring annual bias audits of automated hiring tools. The DPDPA’s 2025 Rules do require Significant Data Fiduciaries to conduct algorithmic fairness assessments, and the proposed AI (Ethics and Accountability) Bill, 2025, explicitly prohibits AI discrimination in employment. But the DPDPA Rules are not yet in full force, the Bill is a private member’s proposal that has not passed, and there is no clear provision today that gives you the right to challenge an automated decision about your employment.
You apply for a job. A machine rejects you. You get a polite email. You move on to the next application. You never know that the machine scored your resume lower because of your name, your college, your city, or the way you phrased your achievements. You never know because nobody is required to tell you.
And here is the uncomfortable truth: the companies using these systems may not know either. They buy the software, plug it in, and trust the output. The hiring manager sees a shortlist of ten candidates out of a thousand. They assume the system selected the best ten. They never see the 990 who were filtered out or the reasons why. A 2025 Brookings Institution study found something even more troubling: when human recruiters screen resumes in collaboration with racially biased AI, people cannot adequately identify and mitigate traces of AI biases that propagate into their decision making. The AI bias becomes sticky it shapes human choices even when humans think they are making independent judgments.
You have a right to know why you were rejected. But do you?
I’m have written a book about exactly this how AI and automated systems make decisions about your life, where accountability disappears, and what we can do about it. If you want to know morea about this book or order a copy, you can do it here: https://akshaywalimbe.com/beyond-bias/