The Limits of AI in Software Engineering: Why Understanding Still Matters
A software company delivers a product to its customers. Then a new government regulation appears, mandating changes to that software by a specific deadline. The company creates backlog items, breaks them into tasks, and developers start coding. The question is: how much AI should they use to write that code?
Three Levels of AI Adoption
Level one. The developer gives the AI access to the task description, the existing codebase, the version control system, and the development database. The developer then asks the AI to create a new branch, write the code change, and open a pull request.
Level two. The company also uses AI to read the new regulation, compare it with the existing system, write the backlog item, and break it down into smaller coding tasks. After the AI has handled the coding work from level one, the company asks it to perform the pull request review, merge to master, write test scenarios, run all tests, and deploy the change.
