25-35% of new code in large organizations is now AI-assisted.
Everyone's shipping faster.
But someone has to review that code. Maintain it. Debug it at 3am.
Those people are burning out.
The Math
I tracked a typical AI-generated pull request:
Contributor time: 7 minutes
Maintainer time: 85 minutes
Ratio: 12x
And when you request changes? They feed your feedback to the AI and regenerate the whole thing. You're reviewing from scratch.
One maintainer told me he's stopped reviewing PRs entirely: "I can't tell anymore which ones are real contributions and which are someone farming GitHub activity for their LinkedIn."
The Security Problem
Radware's threat intelligence team analyzed 500,000 code samples and found "synthetic vulnerabilities" — security flaws unique to AI-generated code.
Key findings:
AI errors are disproportionately high-severity (injection, auth bypass)
"Hallucinated abstractions" — AI invents fake helper functions that look professional but are broken
"Slopsquatting" — attackers register hallucinated package names with malicious payloads
What This Means for Hiring
New interview question: "Walk me through a bug you personally debugged in this code."
If they can't explain trade-offs, they didn't write it.
The developers who thrive won't be the ones who generate the most code.
They'll be the ones who can tell the difference between code that compiles and code that belongs.
Taste scales. Slop doesn't.
Full breakdown in comments 👇
Top comments (1)
See the [full break down]