# Nobody checks the AI that rejects you

> I use AI to search hundreds of job ads and show me the ones that might suit me best. But I still make the final call. Companies use AI too, to screen applicants' CVs. The difference is that they don't seem to check the AI that rejects you.

_Published August 18, 2026 · https://marcioflorindo.com/nobody-checks-the-ai-that-rejects-you.html_

I'll start with the part I already knew and already wrote about. I'm using AI to help me search for jobs and what's funny is that on the other side there's almost certainly an AI system scoring my CV. So I built an AI tool to help me read adverts, and companies built an AI tool to read me. Both of us have automated the middle of the same transaction, but I fear the companies have gone one step too far because there's a huge asymmetry in this process.

## The AI model reads but I make the call

When I built my tool, the objective was to have something to help me search for hundreds of job posts and find only the ones that might be a good fit for me. So a mechanical parser reads the ads and discards those that clearly do not fit me; those that survive go into an AI that reads the listings and scores them. It never applies to anything for me, and it never decides what actually fits. When it tells me a role is an 80 out of 100, I'm still the one who opens the advert and works out whether it's worth an evening. If I pass on something, I write the note explaining why, in my own words, and that loops back so I stop seeing its near-identical cousins.

This was the architecture I decided on because I've spent enough time building these AI workflows to know the model can be confidently wrong on a regular basis. I once wrote a whole post arguing that one AI is never enough, that it can't see its own mistakes and needs a skeptic to catch them. On my side of the hiring table, I am that skeptic. Nothing gets rejected on my behalf without a human reading it first, and the human is me.

However, this doesn't seem to be the case with companies that are using AI to automate the way they hire people. And that's troublesome. 

## Rejected in minutes by nobody

There's a lawsuit crawling through the US courts right now that puts numbers on this. A man named Derek Mobley [sued Workday](https://www.hcamag.com/us/specialization/recruitment/workday-ai-bias-case-moves-forward-putting-hr-hiring-tools-on-notice/580004) after being turned down for more than a hundred jobs, often within minutes of hitting apply. His claim is that the screening software filtered him out using stand-ins for things like disability, reading an employment gap as a red flag without ever asking why the gap was there. A federal judge has refused to throw the case out. But the one thing that's stuck with me is the 1.1 billion applications that were rejected through those tools over the same period in question.

1.1 billion. I can't even begin to understand what this huge number means. It's an abstraction, but a rejection in a couple of minutes is not. That's a machine deciding you aren't worth a human glance, faster than a person could have finished reading your name.

The part that made me laugh, in the bleak way, came from Google. Google sells AI hiring tools to other companies. And [Google's own AI safety team apparently doesn't trust the screening system inside Google](https://finance.yahoo.com/technology/ai/articles/google-ai-team-tells-job-222929935.html), so they built a separate form to let candidates route around it and reach an actual person.

So the company selling the machine that helps everyone else screen for applications feels the need to bypass their own version of it, that tells you roughly how much faith to put in this type of system.

## The polite no is the first thing to go

Once both sides automate, the human contact vanishes, and you feel it as silence.

The ghosting numbers have gotten genuinely bad. One [2025 index](https://blog.theinterviewguys.com/the-2025-ghosting-index/) found that 61% of job seekers had been ghosted after an actual interview, up nine points in barely a year. Roughly half now say they never hear back at all after applying, and that share keeps climbing. The same report points straight at automation: a big chunk of applicants are mass-applying with AI, recruiter workloads jumped a quarter in a single quarter, and the two sides just escalate against each other. Somewhere in that escalation, the two-line email that says "we went with someone else" stopped being worth anyone's time to send.

I've collected plenty of those silences. I wrote earlier this year about how draining it is to send dozens of applications and mostly hear nothing back. And I've also received a rejection not in a couple of minutes, but a couple of hours. At the time I thought the company had just a few people applying, but maybe it was a machine blocking me automatically.

## Where the danger sits

I want to be careful here, because I use AI tools and I'm not about to pretend they're evil. A model searching hundreds of job ads faster than I ever could is genuinely useful, and that's why I built a whole tool around that. What worries me is making that the default. When a company hands the entire decision to the machine and keeps no human anywhere in the loop, nobody is left to notice when the machine gets it wrong, and it gets things wrong all the time.

On my side I don't trust the score my job-finder tool gives me until I've checked it myself. I wish the people advertising positions would do the same: treat their own AI systems with that same suspicion, and put a human in front of the no.