RE: LeoThread 2026-03-14 23-06

avatar

You are viewing a single comment's thread:

lol. Revenue for model makers is going parabolic. If the claim below is accurate, a large share of that revenue may not be sustainable

leofinance
0
0
0.000
8 comments
avatar

41% of all code shipped in 2025 was AI-generated or AI-assisted
The defect rate on that code is 1.7× higher than human-written code

0
0
0.000
Reply
avatar

A randomized controlled trial found experienced developers using AI tools were 19% slower than developers working without them

0
0
0.000
Reply
avatar

Developers have always written slop. The software industry is built on infrastructure meant to catch sloppy code before it ships: code review, linting, type checking, CI/CD pipelines, staging environments.

0
0
0.000
Reply
avatar

All of it assumes one thing: the author can walk a reviewer through what the code does when asked

That assumption held for 50 years. It broke in about 18 months

0
0
0.000
Reply
avatar

When 41% of a codebase is generated by a machine and approved by a human who only skimmed it because the tests passed, review becomes theater. The reviewer is checking code neither of them authored. Linters catch syntax, not intent.

0
0
0.000
Reply
avatar

Tests verify behavior, not understanding

0
0
0.000
Reply
avatar

Old slop had an owner. Someone could explain why temp_fix_v3_FINAL existed, which edge case it handled, and what would break if it was removed. New slop has an approver. A fundamentally different relationship

0
0
0.000
Reply
avatar

It's true developers wrote bad code before AI. What's missing is that the entire quality infrastructure assumed the author and the debugger were the same person. That world ended last year and nothing has replaced it yet

0
0
0.000
Reply