AI is removing the middle class of software engineering

The year is 2020, and you are a software engineering leader. You meticulously maintain code quality and architecture, ensuring thorough code reviews and fostering a healthy development environment. However, one day, you return from vacation to a chaotic codebase. Colleagues had made numerous changes without proper review, introducing denormalized tables, serverless functions, and Kafka without a clear need.

By 2026, you find yourself reviewing 13 pull requests on a normal Monday, each containing extensive AI-generated descriptions. Your team has multiplied changes since your absence, often generating more code in a day than they used to in an entire year. While the initial results may appear functional, the lack of understanding of the underlying complexities leads to severe issues.

Bugs that should have been resolved by the team take months, as the AI-generated code lacks clarity and maintainability. Trained engineers are essential to untangle this mess, as the rapid code generation by AI makes it challenging to comprehend the system's intricacies. As AI continues to advance, technical debt becomes increasingly difficult to manage, and the gap between skilled engineers and those who struggle with AI-generated code widens.