That's fair. I hope they know what they are doing like you think they do.
On topic, I remember a ND dev talking about how it took him 18 months to finish the Museum level in TLOU2. Thats a level with no combat encounters. One real building with two floors and maybe 3-4 large rooms and it took 18 months to iterate and complete. ND actually finished that game in under 3 years despite the hack, the reshoots and delays so they are very productive. I just thought it was interesting just how much iteration every level takes. Thats manual intervention after notes from directors and play testers. I dont know how AI is going to reduce that iteration time, but Metro devs have talked about how using realtime GI has saved them a lot of time that wouldve otherwise been spent on baking lighting and we all know what nanite can do for creating 10 different LODs for every object in the game. I hope Sony devs switch to these realtime GI and reflections next gen for the sake of reducing dev time even if it means a bigger hit on the GPU.
Realtime GI solutions for fast iteration on art asset production and lighting scenes already exist in many engines across the industry. It's not specific to Unreal... like at all. It's not new technology.
Nanite is bleeding edge. But devs already have polished studio workflows for generating and authoring assets for LOD. So while a Nanite-equivalent technology would save them more time in the future, as you even acknowledge, they've already optimised their workflows to be as efficient as possible under the existing development paradigm.