Internet rule: if a comment starts with "So" it's going to be the least charitable version of what you said, haha.
Nah. If I were reviewing something without having been given the instruction "please be truthful, I need to make this perfect" or similar, I would default to encouraging the skill rather than trying to help perfect the output. In doing so I would overlook minor issues that will iron out with experience across multiple works.
If the person was submitting the result to a competition or something and asked my opinion on that specific work, then my approach would be different. Not cruel and hurtful, course not, I'd just mention things that would otherwise be annoying and petty to bring up.
In the case of pre-sapient AI, it won't hurt (and may even be helpful to the relevant dev) to point out where it's tripping over into uncanny valley. There's no benefit to encouragement in this situation as it's not learning a skill, it is trying to replicate the output of one.
