I keep hearing that AI now writes most of the code at some companies. For people actually shipping Python with Claude Code, Cursor or Copilot:
- Roughly what share of your merged code starts as AI output?
- What do you most often have to correct? Bugs, tests, pandas or data issues, dependencies, design?
- Which skills became more important for you, and which became less important?
Context: I work in risk and data and am deciding what to get good at next. Real examples would help more than opinions. Thanks!