Well, I perceive it as more difficult because of my intuitions about how culpability travels up a causal chain.

For example, if someone dies because of a bullet fired into their heart from a gun shot by a hand controlled by a brain B following an instruction given by agent A, my judgment of culpability travels unattenuated through the bullet and the gun and the hand. To what degree it grounds out in B and A depends on to what degree I consider B autonomous... if B is an advanced ballistics-targeting computer I might be willing to call it a brain but still unwilling to hold it culpable for the death, for example. Either way, the bulk of the culpability grounds out there. I may go further and look at the social structures and contingent history that led to A and B (and the hand, bullet, gun, heart, etc.) being the way they are, but that will at best be in addition to the initial judgment, and I probably won't bother.

Similarly, if five people are hit by a trolley that rolled down a track that agent A chose not to stop, my intuitions of culpability ground out in A. Again, I may go further and look at the train switching systems and so on and so forth, but that will be in addition to the initial judgment, and I probably won't bother.

I find it helpful to remember that intuitions about culpability are distinct from beliefs about responsibility.