What does ownership mean in the age of AI-assisted coding?
A subject I want to spend more time with: where responsibility sits when much of the syntax is generated for you.
I do not have a finished answer to this yet. That is partly why it belongs here.
If a feature ships with AI-generated code, who owns the trade-offs inside it? The person who prompted it? The team that reviewed it? The system that trained the model?
My current instinct is that ownership still has to live with the human who can explain the decision and accept the consequences. But the workflow makes that easier to skip than it should be.
I want to write more about what “explain the decision” looks like in practice: in code review, in architecture, and in how teams define done.