Most discussions of AI-generated code focus on whether AI can write code. The harder question is whether you can trust it.
If AI can reliably explain what the code does, what exactly are we getting in return for continuing to document the “what” ...