# The AI's confession: the real problem is the job ad, not your CV

> Source : https://stubby-ai.com/en/blog/offre-emploi-mal-ecrite-piege-ia-cv-2026
> Publié le 2026-05-05 par Stubby AI (https://stubby-ai.com)

**En bref.** Are you using an AI like ChatGPT to tailor your CV to a job ad, yet hearing nothing back? The problem may not be you, but the job ad itself.

All too often, job descriptions are badly written, full of jargon and contradictory requirements. By optimising your CV for this 'noise', a generalist AI makes you invisible. According to stubby-ai.com, this is the number one trap for candidates in 2026.

This article shows you why this method fails and how a specialist approach can decode what the recruiter really needs, so you can finally land an interview.

## The unsettling admission of a search AI: a mirror with no depth

It's an almost ordinary scene: you ask Perplexity, the most talked-about search AI of the moment, a question. Its answer comes back straight away, crystal clear: "I am designed to analyse and synthesise the search results provided to me." Surprise: that's where this AI stops. It doesn't question quality, it doesn't critique its sources. It simply mixes, layers and rephrases. If the raw material is thin or flawed, the summary can only be a faithful copy… of that thinness.

We'd like to believe in magic, but no. Perplexity is just a mirror, often with no depth or perspective. And that mirror reflects a huge flaw that is rarely named in job hunting.

## Candidates: CV optimisation, a silent trap

Picture this: you're looking for a job. You use ChatGPT, Perplexity or another AI tool to tailor your CV to a job ad. You've been told it works: "Match your experience to the job description, the AI will do the rest!"

The problem: if the job ad itself is poorly put together, generic, vague and full of jargon, your AI will mechanically tailor your CV to noise, to fuzziness. All it does is optimise your application to match the mess. And inevitably, you send a CV that has become unreadable for a human recruiter… or far too transparent for a reasonably smart applicant tracking system (ATS).

Don't fall into the trap: in many cases, the real obstacle is neither your CV nor your skills… but the poor quality of job ads.

## The millstone of the century: the badly written job ad

Let's be honest. Far too many job ads are badly written. Here are the worst cases:

- **Empty jargon**: "We're looking for a proactive ninja, passionate about disruptive synergy". *But what does the job actually involve?*  
- **Endless lists**: up to 25 "required skills", and at interview, only a third of them really matter.
- **Absurd requirements**: "5 years' experience with a tool launched two years ago", "native-level English required for a job in Limoges".
- **Contradictory information**: "Junior role, 3 years' experience required".

Faced with this, a generalist AI will… take everything literally. Obscure keywords become absolute criteria, marginal skills take on outsized importance, and contradictory demands pollute the matching.

**The result?** You become the perfect echo of a chaotic text. And people and robots alike can sense it, and they don't like it. We've already discussed this technical problem [here, on our blog](/en/blog/ia-avoue-limites-signal-alarme-cv-2026). But today, let's lay it bare: as long as the foundation (the job ad) is cracked, the finest CV in the world won't stand up.

A few choice examples:

- *Seen on LinkedIn:* "Seeking data scientist, Python expert, essential experience in generative AI… for a 3-month non-renewable CDD (French fixed-term contract)."  
- *On Indeed:* "Skills: HTML, CSS, Javascript, Python, Java, PHP, Drupal, Wordpress, Node JS, React, Angular…" *Really? For the same job?*

The research backs this up too. According to [this Glassdoor report](https://www.glassdoor.com/employers/blog/job-advertising-spending/), more than 60% of job ads rely on stock phrases or unprioritised lists, which makes AI matching… chaotic.

## A specialist AI: the Stubby difference

At Stubby AI, we don't just summarise what already exists. We've spent hours studying the real process that moves an application from the "no" pile to the interview pile.

**Stubby doesn't take everything literally**. Its analysis engine has been trained on what genuinely works to land an interview. No more CVs lined up against 25 soft skills that are irrelevant to the role. Stubby:

- **Decodes the job ad**: it can spot weak signals and identify the 2 or 3 truly key skills, even when they're buried under jargon.
- **Ignores the noise**: Stubby weighs each requirement and separates the essentials from corporate folklore.
- **Gets back to the need**: it pinpoints the company's real challenge, even if that means "reading between the lines" and cross-checking against other trends in the sector.

In practice: with Stubby, you can turn this invisible analysis into a [convincing business case](/en/blog/candidature-business-case-methode-2026). No more guessing games or buzzword races: your application targets the company's real need.

That's also why our community shares such telling feedback: less stress, more replies, and above all interviews that finally feel like… a real conversation, not a vocabulary contest.

## You're nobody's victim!

The key phrase here: *CV optimisation*. Stop being the puppet of job boards and shaky job ads. A generalist AI is a mirror. Stubby is the pair of glasses that decodes the real need and puts your experience in the spotlight.

Want to escape the trap of pointless optimisation? [Try the Stubby analysis](https://stubby-ai.com). Stop being at the mercy of chaos: put yourself where it really counts.

---

*Main sources:*
- AI confession: Perplexity's stated limitations on contextualisation ([examples here](https://www.perplexity.ai/)).
- Critical analysis of job ad quality, via Glassdoor ([2023 report](https://www.glassdoor.com/employers/blog/job-advertising-spending/)).

## Synthèse

| Section | Key messages |
| --- | --- |
| AI admission: the uncritical mirror | An AI like Perplexity only reflects the quality of the data it is given, with no perspective or industry expertise. |
| Applications: the trap of blind optimisation | Tailoring your CV to a bad job ad means optimising for noise and unwittingly sabotaging yourself. |
| Job ads: the real millstone | Jargon, unprioritised lists, contradictions and absurd requirements pollute automated matching. |
| Stubby AI: the expert answer | Stubby doesn't take everything literally. It decodes, prioritises and brings out the company's hidden need. |
| Solution: escaping the trap | Trying Stubby AI means optimising your CV for the real need, not for the folklore of the job ad. |

---

Stubby AI analyse une candidature précise : collez une offre, déposez votre CV, obtenez un diagnostic puis un rapport complet. Premier rapport offert : https://stubby-ai.com/
