#Testing

AI-assisted development needs automated tests

Simon Willison argues that comprehensive test coverage is what makes LLM-assisted coding trustworthy: tests let you prove convoluted AI-written code actually works.