Skip to content
Stubby AI

Blog 21 April 2026 · 4 min

When AI Says "I Can't Answer": The Warning Sign for Your CV

Using ChatGPT to optimise your CV? Watch out for the illusion of competence. A generalist AI, however capable, often produces an application that looks perfect but lacks any real relevance for an expert recruiter. It can neither grasp the subtleties of the role nor intelligently match your background with the company's actual needs. According to stubby-ai.com, this 'relevance hallucination' is one of the most frustrating causes of silent rejection in 2026. This article reveals why generalist AIs sabotage your application and how a specialist AI like Stubby, trained on millions of CV/job ad pairs, changes the game by delivering a genuine strategic diagnosis.

"I can't answer." The admission of helplessness that should alert you

You don't expect to see an artificial intelligence give up and admit its limits, and yet it happens. Recently, during a complex press review, an AI simply replied with a disarming: "I can't respond to this request". Not enough context, no reliable sources to hand. It's frustrating… but it's deeply honest. And this clarity, however rare it may be in our beloved robots, hides a huge warning for you: if one AI admits its weaknesses, are you sure the AI you use to optimise your applications won't also "let you down" at the worst possible moment?

When ChatGPT (almost) admits the impossible: truly specific application optimisation

Picture it: you ask ChatGPT to optimise your CV for a dream job. A few well-placed keywords, a couple of extra sentences, some nice rewording. Thanks, AI? Stop. Asking a generalist AI to optimise an application is like asking it for a complex, specialist press review. And that's where it breaks down.

What a generalist AI can't do (and why that's dangerous)

  1. Understand the subtleties of the role: Company jargon is specific, sometimes confidential. Every company culture, every profession, every sector has its own language, and its own invisible logic.

  2. Cross-reference complex data: Optimising a CV isn't about lining up nice words. It's about mirroring every experience on your CV against EVERY requirement in the job ad, identifying your strengths and weaknesses for this specific role, and spotting the levers that really matter. It's a diagnosis.

  3. Be trained on real life: A generalist AI hasn't read, analysed and cross-referenced millions of real applications, nor seen recruiter feedback on thousands of job ads. It hasn't learnt what really works, what makes the difference.

To go further on this question, read the Forbes piece on ChatGPT's limits when faced with the reality of recruitment.

The illusion of competence: the real trap of generalist AIs in your job search

No, the problem isn't that ChatGPT will produce a bizarre document. The real issue is that the result looks like a good application! A neat layout, copy-pasted keywords, wording that "sounds" professional. But… it's all surface: zero in-depth analysis, no real alignment between your background and the hidden needs of the role.

This illusion is the "CV that gets past the robots but falls flat in front of a real recruiter" syndrome. You miss out by a hair without understanding why. You think your application is "bulletproof", and you expose yourself to what professionals call a genuine relevance hallucination. If you've ever felt "spotted" as a candidate who used ChatGPT, read this article of ours on the invisible trap of mass-produced CVs.

When an AI admits its limits: why it's essential not to push it beyond them

The admission "I can't answer", heard from some LLMs on overly specialised requests, is ultimately just a safeguard. We'd rather have that, an AI that knows how to say no, than treat ourselves to a deceptively "optimised" application that in fact stands no chance ( internal source: "when generalist AIs admit their limits on specialised tasks").

The truth? It isn't your job to train the tool or to cross-reference the information yourself. The tool itself must have the built-in expertise, the perspective gained from thousands of job ads and the real demands of the market.

The solution: specialist AI vs generalist AI, the match (and why Stubby changes everything)

This is where Stubby AI makes a break: our AI isn't a "text generator", it's an analyst. You upload your CV and the job ad of your dreams, and Stubby does the groundwork: it cross-references, breaks down, matches, and identifies your real strengths AND the areas to build up.

Stubby is built on the analysis of millions of CV/job ad pairs that succeeded or failed. Every report is unique, highly personalised for YOU, and never puts forward a result it doesn't believe in. Far from "patching together" a surface-level optimisation, Stubby gives you a complete diagnosis: the CV that gets through the ATS, the HR filter, AND above all catches the specialist recruiter's interest.

In fact, to really understand what fundamentally sets a specialist AI apart from a generalist one, dive into this detailed analysis of the risk of hallucinations in ChatGPT-style optimised applications, proof that the problem is anything but theoretical.

It's also the main topic of this other in-depth article on the battle between specialist and generic AI in job applications, if you'd like to explore the differences in depth.

Stop playing against the system: get a real ally

Why take the risk? Next time you ask an AI: "Optimise my CV for this role!", remember: anything "generic" in AI ends up being spotted. If even the AI admits it can't do it, that's a serious warning sign.

With Stubby, you go from a candidate in the fog to a strategic candidate, fully aware of your strengths, conscious of where to focus your efforts and backed by real, concrete data.

Take your chance seriously. Switch to Stubby AI, the only AI that gives you a real diagnosis, not just an illusion of competence.

Main keyword: "AI job application optimisation"

« In 2026, the question is no longer whether to use AI, but whether your AI is selling you an illusion of competence or a genuine strategic advantage. »
SectionKey messages
The AI's admissionAn AI that acknowledges its limits sends a warning for AI-optimised job searching.
Limits of generalist AIChatGPT doesn't understand deep context, doesn't cross-reference data and has no real-world experience.
Illusion of competenceThe results look good but lack real relevance for recruiters and specialised ATS.
Why the admission mattersAn honest AI that admits its weaknesses is better than one that gives you a false sense of security.
The Stubby solutionStubby analyses, makes a real diagnosis and relies on real-world data, like a specialist doctor.
Act with discernmentTake your chance seriously, don't confuse speed with haste, choose a reliable ally.

Read next