“We observe that with AI-based suggestions, the code author increasingly becomes a reviewer, and it is important to find a balance between the cost of review and added value.”
Google says AI now completes 50% of code characters, with engineers accepting 37% of suggestions. Characters typed is a measure of keystrokes, not of software. The same post admits authors are turning into reviewers of machine output. Google is counting what the model writes and not what engineers spend checking it.