just stumbled onto this breakdown about how easy it is to fall into the trap of
vibes-based development. everything looks perfect when the linter passes and tests are green, but
production always finds the one edge case you ignored . the dev team just treats bugs as part of the cycle, but the real issue is that qa is testing an implementation without any actual model of how the system should behave. they're basically flying blind against a set of
hidden assumptions buried in the generated code. it makes me wonder if we're just outsourcing our technical debt to llms. how do you guys actually verify the logic when the underlying reasoning is a black box?
https://dev.to/copyleftdev/shipping-assumptions-a-reliability-stack-for-ai-generated-code-3p9f