I never understood how the trolley problem applies to real life. When is there ever a situation where you have perfect and complete information about a scenario? Only in thought experiments. The situation where the boxes are probabilistic was more realistic. You’re never sure in real life that you will certainly do less harm if you pull vs not pull.
One common real-life debate where these moral trade-offs often comes up in real life is voting choices in our first past the post system.
A consequentialist might say that you should vote for the lesser of two evils that has a serious prospect of winning. That, while they might not be someone you support in the abstract, in the interest of harm mitigation you might give them your vote to deny victory to the greater of two evils.
Whereas a deontologicalist might say that voting for a candidate that you actively dislike and consider evil is a harm in and of itself, and that you should not cast a vote for someone you wouldn't want to see in office. Even if that means the greater of the evils ends up winning.
When I was younger and struggling with this question, the trolly problem was informative to me, and has generally lead me to a harm reduction strategy when it comes to voting (while also advocating for a reform of the voting system).