Reading and understanding the code is essential, but in a collaborative environment, reviewing AI-generated code can be complex. It's about balancing trust in AI with the need for human oversight.