I made a mistake.

A big one.

I was hiring for a Marketing Manager role, and instead of writing the job description myself like a normal person, I thought: Why not save 20 minutes and let ChatGPT do it?

So I opened ChatGPT and typed: "Write a job description for a Marketing Manager at a SaaS company."

ChatGPT delivered. Two hundred and fifty words of pure, distilled corporate jargon. Phrases like "synergize cross-functional initiatives" and "drive strategic alignment across stakeholder ecosystems."

I copy-pasted it directly into our ATS. Posted it to LinkedIn, Indeed, and our careers page.

Within 48 hours, I had 127 applications.

Great, right?

Wrong.

Every single applicant was also written by AI.

The Applications Were... Suspiciously Perfect

The first red flag: every resume looked identical.

Not literally identical—different names, different companies—but structurally identical. Same section headings. Same bullet point phrasing. Same "achievement-oriented format."

Example bullet points from three different candidates:

• "Spearheaded cross-functional go-to-market strategies, resulting in a 35% increase in customer engagement and a 22% improvement in stakeholder alignment."
• "Orchestrated synergistic brand initiatives across multiple channels, driving a 40% uplift in market penetration and optimizing ROI by 28%."
• "Pioneered innovative digital transformation strategies that enhanced operational efficiency by 33% and elevated customer satisfaction metrics by 19%."

These sentences mean nothing. They're grammatically correct but semantically empty. They sound important but contain zero actual information.

I Googled one of the bullet points verbatim.

It appeared on 47 different resumes—across LinkedIn, Indeed, and random recruiting sites. Forty-seven people had used the exact same AI-generated sentence to describe their "achievements."

The Phone Screens Were Even Worse

I scheduled phone screens with 10 candidates. Surely a live conversation would reveal who was real and who was a chatbot wearing a human name.

Candidate #1: "Tell me about your marketing experience."

"I'm a results-driven marketing professional with a passion for leveraging data-driven insights to optimize customer engagement and drive cross-functional alignment in fast-paced, high-growth environments."

"Okay, but what did you actually DO at your last job?"

"I spearheaded strategic initiatives that enhanced brand visibility and maximized ROI through innovative, synergistic campaigns."

"Can you give me a specific example?"

"Absolutely. I collaborated with stakeholders to deliver impactful solutions that exceeded KPIs and fostered a culture of continuous improvement."

At this point, I realized I was talking to someone reading AI-generated interview responses off a screen.

Candidate #2 accidentally left their microphone on.

I heard them typing. Then I heard a robotic voice—text-to-speech software—reading an answer.

Then they repeated the answer verbatim into the microphone, as if they'd come up with it themselves.

I ended the call.

Candidate #3 was actually ChatGPT.

I'm not kidding.

I asked a simple question: "What's your favorite marketing campaign you've worked on?"

They responded:

"As an AI language model, I don't have personal experiences or preferences. However, I can provide examples of effective marketing campaigns if that would be helpful."

I stared at my screen in disbelief.

Did this person accidentally paste a ChatGPT response instead of rewriting it?

Or—and this is the terrifying possibility—did they just... not realize they'd done it?

I Realized The Problem: My Job Description Was AI Bait

I went back and re-read the job description ChatGPT wrote for me.

It was full of phrases like:

• "Dynamic, fast-paced environment"
• "Drive strategic initiatives"
• "Collaborate with cross-functional teams"
• "Optimize synergies across stakeholder ecosystems"

This job description was algorithmically perfect bait for other AI tools.

When candidates fed my job description into ChatGPT and asked it to "write a resume that matches this job description," it generated resumes that mirrored my AI-written JD word-for-word.

I had created an AI feedback loop.

AI wrote my job description. AI wrote their resumes to match my AI-written job description. AI wrote their cover letters. AI wrote their interview answers.

At no point in this process did an actual human with actual marketing experience participate.

I Rewrote The Job Description (As A Human)

I deleted the AI-generated job description and wrote a new one myself. This time, I used normal language.

Old (AI-written) version:

"We're seeking a dynamic, results-oriented Marketing Manager to spearhead strategic initiatives and drive cross-functional alignment across our go-to-market ecosystem."

New (human-written) version:

"We need a Marketing Manager who can run our content strategy, manage our blog and social channels, and help us figure out why our Facebook ads cost $47 per click when our competitor pays $4."

Old version:

"The ideal candidate will leverage data-driven insights to optimize customer engagement and maximize ROI in a fast-paced, high-growth environment."

New version:

"You should know how to use Google Analytics, write emails that don't sound like a robot, and be okay with the fact that half of marketing is guessing and hoping it works."

I posted the new version.

The Results Were Immediate

Applications dropped from 127 in 48 hours to 31 in the first week.

But—and this is critical—the quality skyrocketed.

The 31 applications I received were from actual humans who had read the job description, understood what the role involved, and tailored their resumes accordingly.

One candidate wrote in their cover letter:

"I saw your note about Facebook ads costing $47 per click. I had the same problem at my last company—we fixed it by narrowing our audience targeting and switching to carousel ads. Happy to explain more if you're interested."

That's a real person. With real experience. Solving real problems.

I hired her.

What I Learned: AI Is A Tool, Not A Replacement

Here's the thing: AI can help you write faster, but it can't make you write better.

When I let ChatGPT write my job description, it gave me something grammatically correct but utterly generic. It optimized for keywords, not clarity. It sounded professional but said nothing.

And because it said nothing, it attracted candidates who also said nothing.

If you use AI to write your job descriptions:

1. Edit it heavily. Remove jargon. Add specifics. Make it sound like a human wrote it.
2. Include details AI can't generate. Specific projects. Real challenges. Actual team dynamics.
3. Ask questions AI-written resumes can't answer. "What's a marketing campaign you're proud of?" is easy to fake. "Why did our Facebook ads cost $47 per click?" is not.

If you're screening candidates who might be using AI:

1. Ask follow-up questions. AI-generated answers don't hold up under scrutiny.
2. Request work samples. AI can write bulleted lists. It can't create a real marketing campaign from scratch.
3. Listen for jargon overload. If every sentence includes "synergize," "stakeholder alignment," or "strategic initiatives," they're reading from a script.

The Bottom Line

I let ChatGPT write my job description to save 20 minutes.

It cost me two weeks of interviewing robots.

Lesson learned: AI is great for drafting. Terrible for thinking.

Use it as a starting point. Not the finish line.

And for the love of God, if you're a candidate using AI to write your resume—at least edit out the part where it says "As an AI language model, I don't have personal experiences."

We notice.