I'm sorry to bring it to you, but just because you're a perfect god at the game, doesn't mean everyone is. I have a total of close to 700 hours in Civ V and VI, if I put a unit somewhere and it gets killed while I didn't expect it (as opposed to a standard casualty of a unit getting fired upon for a few turns and me realizing I can't save it), then it's because the AI is better than I expected. Which, in turn, means it's better than it was in V and previously in VI.
If you're learning the game, it's reasonable that you will lose a unit unexpectedly.
If you are repeatedly putting units in places they die to concentrated fire...why are you doing that? In a strategy game the expectation should be to alter behavior in favor of strategies that actually work. Unless suiciding your units confers benefits, you should probably stop doing it instead of ignoring evidence that you can stop doing it.
U_Sun is a very good player that has massacred civ 4 deity, when beating deity meant that you avoided major mistakes. But he's a human being utilizing a decision heuristic + doing computations while going through the game. There is nothing, in principle, preventing you from doing similarly and going games without losing units.
With hundreds of hours in a strategy title, the ideal is to improve. I know most players don't, and instead sit in a state of perpetual intermediacy (after taking years off from civ 5, I returned briefly and caught up to my playing group inside a week

), but that's *not* a required outcome.
As for the AI, 5 > 3, but 5 is still pretty low if the max is 100. Where 6 is so much more forgiving is that it simplifies the planning and computation for micromanagement. All you need is a vague understanding of where to slap down districts, enough cities, and to know how to survive and you've a good chance of winning if you know the game rules. There are less early game decisions to juggle in the more recent entries in civ.
I'd have less issue with that if the game just let it play through faster then (fewer, but very strong implication decisions can be a decent design approach), rather than adding heaps of decisions that don't actually alter the outcome much but burn time to slog through them.
Considering it can't put 5+ people in MP in stable fashion, has serious design limitations on meaningful decisions, a weak AI due to the game constraints, and terrible UI 6 has no business being considered for best in the series, because not all of the earlier titles share those flaws.