今日の宿題は難しかったと見えて、誰も終わらせてなかった。
The given translation is
One can conclude that today’s homework was difficult, because no one completed it.
But I feel that the causal relationship should go in the opposite direction, i.e:
Because today’s homework seemed difficult, no one completed it.
I feel this way because AてB can mark A as a reason for B.
Am I misunderstanding something here?
The translation is correct. I can’t explain it myself, but re-reading the grammar point makes it pretty clear how this grammar functions:
This use of the と particle in Japanese is called a result, and marks when (A) is something that has come to a conclusion, or can be concluded. In this way とみえる literally means ‘from the way it appears, one can conclude that it’s (A)’. However, this translates a bit more smoothly as ‘it can be deduced that (A)’, or ‘it seems that (A)’.
This expression is always used directly after the phrase that it is drawing as a conclusion.
と always ‘sums up’ things. As a quotation, it sums up the phrase before it as either something that was said, or thought. As 一緒 ‘being together’, it sums up groups of things that are existing/performing actions mutually. Lastly, as a result, it sums up causes, and then illustrates their effects.
So it’s not “it looked difficult and because of that nobody finished it” but more “We can conclude this homework was difficult; nobody finished it”.
Thanks for posting it as it also made me revisit a quite old grammar point that I didn’t previously pay so much attention to 
I guess the part that confuses me the most is that with 「A」と見えて、「B」the conclusion (A) comes before the observation (B) that it stemmed from. There are similar examples of this in the grammar point that I’ve missed:
彼が今プールから出て来たと見えて、全部が濡れている。
He is completely soaked. It can be deduced that he just got out of a pool.
Here the logical conclusion (“he just got out of a pool”) goes BEFORE the observation/fact (“he is completely soaked”). This might strange at first but after I thought about it, it is actually also quite common in english:
He must have just gotten out of a pool, judging from the fact that he is completely wet.
Yes exactly that. Nothing says the observation has to come first, even in English. But even if in English that was the order, Japanese isn’t English
Japanese is also very flexible with the order of many things, so you shouldn’t be trying to find very strict rules in that
と見えて is not a very intuitive turn of a phrase for sure.
In the end it’s just a construction to remember:
Aと見えて、B。
Bだから、Aなんだと思う。
But maybe it’s helpful to think about this て as describing the clause following it:
今日の宿題は難しかったと見える。
it looks like today’s homework was difficult (conjecture, basis for it is not stated)
今日の宿題は（難しかったと見えて）誰も終わらせてなかった。
in a way that makes it look like it was difficult (same conjecture), nobody completed today’s homework (fact providing the basis).
There’s a bit more discussion about it:
Of course, Bunpro’s article should expressly describe what 「Aと見えて、B」 means, instead of sneaking it into a test sentence for 「Aと見える」 and expecting the learner to figure it out.
Now I have some more insights new this - I might be wrong, but that’s how I mentally reconcile with this construct:
The two statements “今日の宿題は難しかったと見える - we can say that today’s homework was hard” and “今日の宿題は誰も終わらせてなかった - no one finished today’s homework” exists as a logical duality that infers each other, namely:
The inference goes both ways and both parts have to appear as a pair. So, if we drop all the causality markers and the assumption that causality must have a direction, we have two sentences that are with almost identical meaning:
… albeit with slightly different emphasis.
I think another thought experiment to do is to try saying:
今日の宿題は誰も終わらせてなくて、難しかったと見える。
Would it change the meaning of the sentence? Or … is it even correct Japanese?