AI demands more engineering discipline. Not less
Charity Majors clarifies she is not telling anyone to skip code review. She traces how AI code generation went from 'slop' to median-engineer quality with Opus 4.5 in late 2025, making code production nearly free. The real product of a team is shared understanding, not lines of code. The post does not spell out specific new disciplines she recommends.
Why it matters: Charity Majors traces AI coding from early-2025 junk to Opus 4.5 at average-engineer level, arguing shared understanding matters more than lines of code.