I'll go along with Qxa8 wins a rook right now, as opposed to one move later. But you're about to discover something about chess engines, which will be useful later on, no matter how good you get at chess.
The line the intrigues me is 8. Qxa8 Nc6 with the idea of trapping the Q out of play for a while. What can Black get from that? So perhaps at this point, when the engine suggests Qxa8, explore what it is the engine sees by going ahead and playing it and then playing a response that might make sense, and see what the computer wants to continue with.
Then, once you think you've milked everything you can learn about the computer's suggestion, go back to the position and make your move (8.Bd3) and see what the computer wants to play against it. (8. ... g6 isn't necessarily the only response you'll want to explore, btw. You could use your engine to explore the ramifications of 8. ... f5 while you're at it; will the king flight through f7 work well enough to save the rook? I'll leave that as a fun question for you to answer on your own.)
While you do that, you may discover than one of the most useful things about an engine is they never get tired of playing different moves from a position. So when you get to a position where you disagree with the engine, you can play the position out multiple times with multiple responses; the engine will patiently show you what it's seeing for each of those lines. And the more different continuations you play out, the more you will understand what's going on in the position, and the better you will know what to play when positions similar to it arise in your future games.
The object, after all, is not to learn how the engine works, but to understand chess; and the engine can be a useful tool for that. Use it to explore the position and you can find out.