I think this explains why the current push is towards “just trust the LLM bro, it hardly ever makes mistakes anymore bro, you don’t even need to be able to read the code anymore”. Converting time spent writing code into the same amount of time reviewing code just makes everyone doing the work miserable and doesn’t eliminate jobs the way management wants. So both camps have an incentive to say it’s fine to just trust the output now https://arstechnica.com/ai/2026/10/ai-coding-agents-generate-more-code-but-not-more-software/