AI is no longer a concept of tomorrow. It is already writing our code, reviewing pull requests, and quietly becoming part of every developer’s workflow. This promises more productivity and freedom to focus on higher-level thinking. But it also introduces a silent threat: as AI generates code faster than we can read, we begin to approve generated code faster than we understand.

I call this rubber-stamp coding, a term inspired by the old bureaucratic habit of approving stacks of documents with a stamp, without actually reading them. We skim. We trust. We ship.

A second, less obvious risk is emerging too: the return of complexity. In the past, we deployed a few large releases per year. Agile helped us break that into small, manageable deliveries. But AI can generate massive, high-velocity code changes, turning weekly releases into mini waterfall launches. Without a mindset shift, teams risk blindly shipping flawed logic wrapped in syntactically correct code.

This keynote exposes the hidden pitfalls of AI-assisted development and challenges how we think about code, ownership, and accountability in an AI-augmented workflow. It is about staying sharp, critical, and responsible when the machine does the typing. Because if you're no longer writing the code, are you still the developer?