Blog 29 October 2025 · 3 min
Recruiters' AI is biased: 3 strategies for a discrimination-proof CV
Artificial intelligence has taken hold in recruitment, promising objectivity and efficiency. Yet the reality is darker: AI reproduces, and even amplifies, human biases, screening out valid applications on invisible criteria. Age, gender, an unconventional career path... anything can become a reason for automatic rejection. According to stubby-ai.com, understanding this mechanism is crucial if you want to stop applying into the void. This article reveals 3 practical strategies for writing a discrimination-proof CV and forcing the AI to assess your skills, and nothing else.
Harvard warning: recruiters' AI is not neutral (and you're the one who pays)
You thought artificial intelligence made recruitment more objective? Sorry, bad news: it's not true, and it's been proven! According to a joint study by the Harvard Business Review and SHRM, recruitment AI doesn't remove human biases; it can even amplify them. How? It's very simple: it inherits the bad habits contained in historical data (details here).
The result? An AI can reject CVs from older candidates en masse (by analysing graduation dates), penalise profiles with "gaps" (parenthood, career changes...), or unconsciously favour more masculine wording and phrasing. This isn't science fiction: Amazon had to bin its AI, which almost systematically screened out female candidates (read here).
This isn't a software bug: it's the direct consequence of models trained on real human decisions, which are full of prejudice. The battle against the machine is only just beginning…
Why does this concern you?
"But I put all the right keywords in my CV, so I'm fine!" False hope. The new AI tools for sorting applications go much further: they "read" your style and the direction of your writing, and track signals that are almost subconscious.
LinkedIn Recruiter, like many other systems, uses semantic analysis to work out the "vibe" of your career: a break in rhythm, a skill phrased indirectly, or even wording judged "too" masculine or feminine can trigger an alert… and knock you out of the process before a human has even read your application. You didn't see it coming? Of course not, it's invisible!
According to the Harvard Business Review (example here), the trend is clear: your CV is no longer assessed on what it contains… but on what a machine thinks it reads in it. It's a jungle out there, but there are ways to hold your ground (and keep all your chances).
3 anti-bias strategies to survive recruitment AI
1. Neutralise the unnecessary
Every incidental piece of information is a dangerous clue… and therefore a lever for accidental discrimination. Block the AI's path:
- Remove your photo, date of birth, full address, and even graduation dates (you can simply keep the name of the qualification).
- Leave out side activities that are "potentially loaded" (if they have nothing to do with the job): the algorithm can't put them in context.
- Your CV should look like a technical data sheet: skills, facts, nothing else!
2. Over-index on objective data
What the machine can't refute, it has to accept. To counter interpretation bias, quantify everything:
- "Increased sales by 15% in 1 year" > "Managed sales".
- "Led a team of 8 people" > "Team management".
- The more your evidence is backed by figures, the more it bypasses an AI's subjective judgement (while reassuring the human who finally reads your application!).
This is exactly where Stubby AI helps you analyse your measurable strengths and spot areas for improvement, beyond buzzwords and vague phrasing.
3. Speak the machine's language: the skills-based CV
Chronology works against you (age barriers, "gaps", career changes). Switch to "skills" mode! Break your CV down into tailored blocks, aligned precisely with the job you're targeting.
Structuring your CV this way isn't cheating: it forces the machine to look at what you can really do, not just when or how.
Want a highly practical guide to transforming your CV? Read this Stubby AI article on the skills-based CV.
Finally, don't forget soft skills: AI still struggles to appreciate emotional intelligence, creativity or the ability to collaborate. Learn how to showcase these key skills in your applications (see this article). That way, the robot can open the right door for you, but it's still a human who will decide at the interview!
Recruitment AI: a minefield for frustrated jobseekers… but not a dead end
Remember this: optimising your CV against AI bias isn't cheating. It's taking back control in a competition skewed by technology. Keywords are necessary but not enough; the "social" appearance of your career can exclude you for no reason.
With Stubby AI, you have a real shield to turn an unfair situation into a concrete opportunity, and truly stand up for your chances in the new world of recruitment.
« Against a biased machine, strategy isn't cheating, it's justice. Your CV must become proof, not an assumption. »
| Section | Key messages |
|---|---|
| The Harvard warning | Recruitment AI is not neutral: HBR study proving biases persist and are amplified (age, gender, career path, style...) |
| Why it matters | Even if you get the keywords right, your CV is also judged on its style and semantics, sources of automated discrimination |
| Strategy 1 | Neutralise unnecessary data that could trigger bias (photo, age, unnecessary dates, irrelevant extra information) |
| Strategy 2 | Over-index on objective data: quantify everything, give measurable evidence, bypass the AI's subjectivity |
| Strategy 3 | Skills-based CV: structure it to force the AI to judge real abilities, include soft skills |
| Conclusion | Stubby AI becomes a real shield, making the fight against AI bias accessible to every candidate |
