There's not difference in maintainability in AI vs Human generated code

I get the best experience with AI when I treat it as another human being, providing guidelines, principles to follow and the right context to operate in. LLMs are great at pattern recognition, so as long as the codebase is well designed and architected, it'll produce high-quality output (for my standards at least)

Overall, we did not detect systematic maintainability advantages or disadvantages when other developers evolved code co-developed with AI assistants. - Echoes of AI - Investigating the DownstreamEffects of AI Assistants on SoftwareMaintainability

• See Context compression for better code generation for how Netflix works with AI
• Also that's why Software systems integrating with AI still benefits from good software architecture