AI-generated code is a ticking time bomb, warns Lightrun's Moshe Sambol. The enthusiasm among managers to adopt AI tools has outpaced developers' ability to learn those tools and use them effectively, creating a situation of 'pain waiting to happen'. Sambol highlights the varying degrees of AI tool adoption, from organizations that have banned developers from writing code to those just starting to roll out AI due to compliance obligations. The pressure to be more productive is immense, but not everyone manages to keep up. Sambol expresses sympathy for developers who have been directed to use AI tools without proper training and organizational guidance. He warns that generative AI models will produce a lot of code quickly, and because the code seems correct initially, it often gets pushed forward, potentially creating bugs and technical debt. Sambol emphasizes the importance of developers being able to explain and validate the code they write, especially in large projects where no one person understands the entire system end-to-end. He describes an incident where a developer used an AI assistant to build an Ansible automated workflow, which ultimately caused the service to stop working due to the AI model not remembering its previous guidance. Sambol points out that studies show a significant percentage of AI-generated code contains errors and creates technical debt. While acknowledging that developers have their own weaknesses, Sambol argues that it's crucial to acknowledge imperfection and work towards processes that improve results, such as automating the prompting process to make it more repeatable.
