Why AI Can Write Code But Can’t Keep It Running (Yet)
AI exceeds expectations at writing code but struggles to handle operational complexity of running systems
“The question, I think, is complexity and the extent to which you can take the operational things you describe and turn them into something that AI can understand”
An a16z podcast conversation explores the gap between AI's strong code-generation capabilities and its unproven ability to manage operational complexity—running, monitoring, and maintaining distributed systems over time. The core argument is that training data for operations is harder to formalize than code, making it an unsolved frontier. This is a useful framing for practitioners but lacks concrete evidence or a named breakthrough.