> If a calculator works great 99% of the time you could not use that calculator to build a bridge. But if the alternative is doing calculations by hand (writin

> If a calculator works great 99% of the time you could not use that calculator to build a bridge. But if the alternative is doing calculations by hand (writing code manually) there is a higher chance of making mistakes. Just like calculations are double checked while building bridges unit tests and code reviews should catch bugs introduced by LLM written code.