Ryo Suwito

Posted on Dec 19

Buy More Ramen, Devs! Because for Years Ahead, AI Slop ATS Will Kick Our Godly Resumes Hard in the Pants

#ai #webdev #programming #career

The Suspicious Interview

So as a seasoned 7-year vibe-coder, recently I've been noticing some interviewers been "busy" on the cam...

Like, they're typing something, moving shit around, eyes looking here and there...

LIKE MAN, I've been on countless interviews before. Mostly they just come with their head and start yapping. But now? They seem to be using "something" suspiciously to help profile myself and other candidates in real-time.

That got me thinking... and what I found is way worse than I imagined.

The Hypothesis: Something's Fundamentally Broken

I had this shower thought: What if the hiring decline we're seeing isn't about "talent shortage" at all? What if companies are just... lazy as fuck with their AI implementation?

So I decided to gather receipts. Real data. Not vibes—actual statistics from industry reports, academic studies, and company surveys.

Buckle up. This rabbit hole goes deep. 🕳️

Part 1: The Numbers Don't Lie—Hiring Is Collapsing

Application-to-Interview Ratios Are in Freefall

Let's start with the obvious: it's gotten exponentially harder to even get an interview.

2016: 15.25% of applicants got interviews¹
2023: 8.4% of applicants got interviews²
2024: 3% of applicants get interviews³

That's a 34% decline in just 5-7 years. For every 100 applications you send, only 3 result in interviews.

But wait, it gets worse.

Interview-to-Offer Ratios Are Also Broken

You'd think if companies are being MORE selective at the application stage, they'd have better conversion at the interview stage, right?

Wrong.

The interview-to-offer ratio used to hover around 16-19%⁴
It's now jumped to over 36%⁴—meaning companies need MORE interviews per hire
Meanwhile, the interview-to-hire conversion rate dropped to 27% in 2024³

Companies are interviewing more people but hiring fewer of them. The system is less efficient at every stage.

Time-to-Hire Is Exploding

2023: 36-44 days average time-to-hire⁵
2025: 68.5 days average time-to-hire⁵

That's almost double. And 60% of companies reported increased time-to-hire in 2024⁵.

So let me get this straight: Companies are using AI to make hiring "faster and more efficient," but hiring is taking twice as long as it did before? 🤔

Part 2: The AI Screening Explosion (And It's Getting Worse)

The Adoption Numbers Are Staggering

Here's where it gets interesting:

AI use in hiring doubled from 26% (2023) to 53% (2024)⁶
99% of Fortune 500 companies now use AI-based hiring tools⁷
83% of companies plan to use AI for resume screening by 2026⁸
70% of companies are already using AI screening⁹

But here's the kicker—most of this is ChatGPT/LLM-based screening:

49% of businesses used ChatGPT in hiring in 2023¹⁰
Rose to 53% by 2024¹⁰
93% of those businesses plan to expand its use¹⁰

The Lazy Implementation: Copy-Paste into ChatGPT

And here's how they're actually using it. This is literally documented in multiple recruiting blogs and guides:

Recruiters are:

Copy-pasting your entire CV into ChatGPT¹¹
Copy-pasting the job description¹¹
Asking ChatGPT: "Rate this candidate 1-5" or "Does this candidate match?"¹²¹³
Rejecting based on the output

I'm not making this up. Here's an actual example prompt from a recruiting guide¹²:

"You are an AI recruiting assistant. Below is a list of resumes, followed by the requirements for a Project Manager role. For each candidate, summarize their relevant experience, key skills, and qualifications. Then, rate their suitability for the role on a scale of 1 to 5, with 5 being a strong match."

That's it. That's the "sophisticated AI screening" we keep hearing about.

Part 3: The Frozen Knowledge Problem (Why You're Getting Rejected)

Here's the thing that's absolutely infuriating about this lazy implementation:

LLMs Treat Experience as Snapshots, Not Trajectories

Let's say you have:

5 years of PHP experience
3 years of backend architecture
1 year with Laravel specifically

The job posting says: "3 years Laravel experience required"

Human recruiter logic:
"This person has 5 years of PHP and solid backend experience. Laravel is just a PHP framework—they obviously know it well. Let's interview them."

ChatGPT logic:

Candidate Laravel experience: 1 year
Required Laravel experience: 3 years
1 < 3
RESULT: NOT A MATCH. REJECT.

The AI cannot reason about:

Transferable skills
Framework similarities (Laravel IS PHP)
Career trajectory
Learning curves
Adjacent experience

Your knowledge is frozen in time to the AI. It sees "1 year Laravel" and stops thinking.

This is documented in academic research. A University of Washington study analyzing over 3 million resume-to-job comparisons found that LLMs fundamentally cannot assess career progression or skill transfer¹⁴.

Part 4: The Receipts Are DAMNING

Companies Admit They're Screening Out Qualified Candidates

Here's where the hypocrisy becomes crystal clear:

88% of employers admit that ATS systems screen out highly qualified candidates⁸
Yet 74% of employers claim they're "struggling to find skilled talent"¹⁵
69% of organizations report significant difficulties filling positions¹⁶

Let me spell this out: Companies are rejecting 97% of applicants, admitting they're screening out qualified people, and then complaining about talent shortages.

The cognitive dissonance is chef's kiss 🤌

The Bias Problem Is Severe

A University of Washington study examined how LLMs rank resumes¹⁴:

LLMs favored white-associated names 85% of the time
Female-associated names only 11% of the time
Never favored Black male-associated names over white male-associated names

Not sometimes. Never.

And this was across 3+ million comparisons using state-of-the-art LLMs from multiple companies (Mistral AI, Salesforce, Contextual AI)¹⁴.

The Cost Is Staggering

So what happens when this broken system lets the wrong candidates through?

Mis-hires cost the US economy over $1 TRILLION annually¹⁷
48% of businesses spend $5k-$10k in direct costs per bad hire¹⁸
Bad hires cost an additional $30k-$150k+ in indirect costs (training, lost productivity, team morale)¹⁸
The estimated total cost ranges from 5 to 27 times the person's actual salary¹⁷

And the turnover?

22% of new hires leave within the first 45 days¹⁹
43% leave within the first 90 days¹⁹

Hiring Managers Have No Confidence

Even the people using these systems don't trust them:

50% of HR professionals lack confidence their hiring process identifies the best candidates¹⁸
Only 25% feel highly confident in their organization's ability to gauge quality of hire²⁰
Only 41% say skills assessments are effective despite 82% using them¹⁹

Part 5: The Hypocrisy Olympics 🏆

Companies Use AI to Screen, Reject Candidates Who Use AI

This is my favorite part:

Companies:

97% of Fortune 500 companies use AI for hiring²¹
70% use AI screening without human oversight²²

Also Companies:

46% of recruiters would disqualify a candidate for using AI on their resume²³
19.6% would reject an AI-generated resume outright²⁴
14.5% believe AI shouldn't be used by candidates at ANY stage²⁴

So it's fine when they use AI to filter you out, but it's a "red flag" when you use AI to pass their filters?

The double standard is magnificent.

The AI Arms Race

And here's where it gets absurd:

About half of job candidates are now using AI to write applications²¹²⁵
This creates "higher volume and lower quality" according to recruiters²⁵
So companies deploy MORE AI to filter out AI-written applications
Which means candidates need to use MORE sophisticated AI to get through
By 2028, experts predict 1 in 4 candidates will be fully AI-generated²⁶

It's AI fighting AI, while actual qualified humans get caught in the crossfire.

Part 6: The System Is Getting WORSE, Not Better

More Automation, Less Human Judgment

By 2025, here's what's planned²²:

83% of employers will use AI for initial resume reviews
69% will use it for assessing qualifications through analytical tools
19% will conduct interviews through AI
70% currently rely on AI to automatically screen out applicants

Some companies are already there. Genpact reported that 40% of their hires go through a "touchless process" up to the interview stage¹⁰—meaning candidates never interact with a human until the final interview.

The Slowdown Paradox

Remember how AI was supposed to make things faster?

Organizations struggle with proper AI configuration, leading to SLOWER screening⁵
When AI systems reject qualified candidates or flag false positives, humans spend MORE time correcting mistakes than they would have spent reviewing applications traditionally⁵
Decision-making paralysis affects 81% of hiring managers⁵

The "efficiency gains" were a lie.

The Feedback Loop of Doom

Here's the vicious cycle we're in:

Companies use lazy AI screening (copy-paste to ChatGPT)
AI rejects qualified candidates with non-linear paths
Companies complain: "We can't find talent!"
Companies lower standards OR rush to hire
Quality of hire drops, bad hires increase
Turnover spikes (43% within 90 days)
Companies blame "talent shortage"
Companies buy MORE AI tools
Repeat from step 1 🔄

Part 7: They're Doing It During Interviews Too

Remember my observation at the start? Those interviewers typing away, looking distracted?

They're using ChatGPT during the interview.

Multiple recruiting guides now recommend this:

"Copy the candidate's CV and your notes into ChatGPT during or immediately after the interview"¹¹
"Ask it to identify red flags and areas of concern"¹¹
"Use it to generate follow-up questions in real-time"²⁷

That explains the multitasking. They're not taking notes—they're feeding your responses to an LLM and asking it whether to hire you.

Part 8: What Can We Actually Do?

The Brutal Truth

The system is broken, and it's getting worse before it gets better. Here's the reality:

You can't opt out. 83% of companies are using this by 2026.
The bias is baked in. No amount of "resume optimization" fixes systemic bias.
Quality will keep declining. Bad hires → more AI → worse screening → more bad hires.

Survival Strategies (The Ramen Math)

Given that the average time-to-hire is now 68.5 days and you need to send 500+ applications for a single interview, here's what actually works:

1. Network > Applications (By a Mile)

Referrals have significantly higher conversion rates²⁸
Employee referrals are the highest quality source of hires³
Companies know their AI screening is broken, so internal referrals bypass it

2. Target Smaller Companies

Only 37.5% of hiring managers use AI screening—it's concentrated in large enterprises²⁴
45.2% of Gen X and 72.5% of Boomers don't use ANY AI tools in hiring²⁴
Smaller companies, older hiring managers = better odds

3. Skills-Based Portfolios

98% of employers agree skills-based hiring is more effective than resumes²⁹
Live projects, GitHub repos, deployed apps—show don't tell
If they can see your work, the AI can't reject you for "insufficient years"

4. Use AI to Beat AI (If You Must)

If they're using ChatGPT to screen, use ChatGPT to optimize
Tools exist to analyze job descriptions and tailor resumes
But know this perpetuates the arms race

5. Ask About Their Hiring Process

"Do you use AI screening?"
"What's your human involvement in the process?"
If they're 100% automated, you probably don't want to work there anyway

The Long Winter

Here's the uncomfortable truth: This problem has no quick fix.

AI adoption is accelerating (83% by 2026)
Implementations are getting MORE automated, not less
The people buying these tools don't understand how they work
And there's a trillion-dollar industry telling them "AI will solve your hiring problems"

Average time-to-hire: 68.5 days.
Applications per interview: 333.
Months of runway you'll need: Calculate accordingly.

Stock up on ramen. This winter is long.

TL;DR (The Summary Version)

Application-to-interview rates collapsed from 15.25% (2016) to 3% (2024)
53% of companies now use ChatGPT for screening—mostly by copy-pasting CV + job description and asking "does this match?"
LLMs treat experience as snapshots, not trajectories—"1 year Laravel" ignores "5 years PHP + architecture experience"
88% of employers admit they're screening out qualified candidates
Yet 74% claim they can't find talent (while rejecting 97% of applications)
LLM bias is severe: 85% favor white names, 11% female names, 0% Black male names
Bad hires cost $1 trillion annually, with 43% of new hires leaving within 90 days
Time-to-hire doubled from 36 to 68.5 days despite "AI efficiency"
46% of recruiters reject candidates who use AI on resumes, while 97% of Fortune 500 use AI to screen
By 2028, 1 in 4 candidates predicted to be completely AI-generated

The system is broken. It's not you—it's a critical thinking shortage among people implementing AI.

Buy ramen. Network hard. This is the new normal.

References

Discussion

What's your experience been? Are you seeing this in your job hunt? Drop your horror stories below. 👇

And if you're a hiring manager reading this: please, for the love of god, add human judgment back into your process. Your AI is rejecting people you actually want to hire.

Tags: #career #hiring #ai #recruiting #jobs #developers #layoffs #tech

Withe. (2024). 50+ Job Interview Statistics [2024] ↩
Jobvite. (2025). 7 Benchmark Metrics to Improve Your Recruiting Funnel ↩
CareerPlug. (2025). Recruiting Metrics Benchmarks ↩
Jobvite. (2025). 7 Benchmark Metrics to Improve Your Recruiting Funnel ↩
The Interview Guys. (2025). State of the Hiring Process in 2025 ↩
HeroHunt AI. (2025). How to Use ChatGPT in Recruiting (2025 Expert Guide) ↩
MIT Law. First Come, First Hired? ChatGPT's Bias for The First Resume It Sees ↩
Enhancv. (2023). 170+ Must-Know Resume Statistics for Job Seekers in 2026 ↩
MIT Law. First Come, First Hired? ChatGPT's Bias for The First Resume It Sees ↩
Manatal. ChatGPT Recruiting in 2025: Use Cases, Prompts & Best Practices ↩
Uplers. (2024). AI in Hiring: How LLMs Enhance Candidate Screening and Quality ↩
Manatal. ChatGPT Recruiting in 2025: Use Cases, Prompts & Best Practices ↩
Northreach. (2023). Supercharge recruitment with ChatGPT & Large Language Models (LLMs) ↩
University of Washington. (2024). AI tools show biases in ranking job applicants' names ↩
Manpower Group. (2024). Talent Shortage 2025 ↩
CSS Tec Staffing. The State of the Talent Shortage in 2025 ↩
Rhythm Systems. (2025). How to Avoid the Staggering Cost of a Mishire ↩
Toggl Hire. The True Cost of a Bad Hire: 2025 Report ↩
SuperAGI. (2025). The Future of Talent Acquisition: How AI Skill Assessment is Revolutionizing the Hiring Landscape ↩
Compono. (2025). Full Research: Latest HR & Talent Statistics (2025) ↩
Entrepreneur. (2025). Employers Can Tell If You Used ChatGPT to Write Your Resume ↩
Resume Builder via HR Grapevine. (2025). Employers increasingly turning to AI to handle recruitment process ↩
Fairygodboss. Recruiters - are these chat gpt statistics true? ↩
TopResume. (2025). Survey: Where Employers Draw the Line on the Use of AI in Hiring ↩
TechSpot. (2024). Recruiters overwhelmed as 57% of young applicants are using ChatGPT for job resumes ↩
Final Round AI. You are not getting a job offer in 2025 and the Reason is AI ↩
iSmartRecruit. (2024). How ChatGPT Helps Recruiters in Recruitment? ↩
Ashby. (2024). Offer Acceptance Rates | Talent Trends Report ↩
TestGorilla. (2023). 50 Key Resume Statistics ↩

DEV Community