On the desktop with RTX 4070 Desktop, it takes 2.5h to fill 64 GB RAM. Since the GPU is the bottleneck, it does not depend on RAM details. Therefore, in a notebook with RTX 4080 Laptop, it should also take roughly 2.5h to fill 64 GB RAM so my usage would be the same and I do need those 64 GB while 32 GB for would insufficient for every demanding AI query and 48 GB would insufficient for every second demanding AI query. Therefore, I will never buy a notebook for AI without at least 64 GB RAM.
With 32 GB, I could only do non-demanding AI queries. These amount to ca. 95% while only ca. 5% are demanding AI queries. However, for my study purpose, I do need the full view including the demanding AI queries. Otherwise, there would be serious gaps in understanding the study field whenever things become particularly interesting.
***
When somebody says he needs 64 GB for RAW photo editing or advanced video editing, I believe him. It would not be my need because my media editing needs are only ordinary and partially even basic. 64 GB is for advanced endusers with advanced needs but we do exist and have such needs, regardless of much clown downtalk you spend.