5 ms·

People love to talk about trolley problems, but they almost never happen. Every accident myself or anyone I know of has gotten into was entirely the result of s

by rictic 8y ago

People love to talk about trolley problems, but they almost never happen. Every accident myself or anyone I know of has gotten into was entirely the result of simple human error, with no difficult ethical decisions involved.

nkoren 8y ago
It's a more generalised problem than you probably think, and is a factor in virtually every multi-party accident. Even if the root cause of the accident is someone else's human error, this still leaves surrounding vehicles with a choice about how to respond to it: whether to take damage or inflict damage, and if the latter, then inflict damage upon whom. For an AV, that's a trolley problem. For a human, it's not. We don't rationalise about such things: it occurs in an eyeblink, and gets covered under "shit happens". We convince ourselves that the outcome is inevitable, just a function of physics. But it isn't. Unlike humans AVs will have the ability and obligation to make choices in such situations; this is critical for their harm-reduction capabilities. When all of those choices are to some degree bad (which is very, very often), then it's a trolley problem. Which is why people who work on AVs are legitimately concerned with them.
- xapata 8y ago
  Eh. Or you can just let the auto-driving auto-mobile take the same "shit happens" approach that humans do. Whatever the code happens to spit out, that's the direction it goes, ethics ignored.
  - nkoren 8y ago
    Yep. All that stands in the way is every personal injury lawyer, class-action lawyer, consumer interest group, public safety regulator, competitor, and short-seller in the world. Shouldn't be a problem.
    - xapata 8y ago
      I'll believe that when I see it.
mmt 8y ago
I could be wrong, but my understanding is also that fatalities per event tends to be in the low single digits. There are just a high number of fatal collisions. That, of course, makes sense if cars generally have 1 or 2 occupants and accidents are usually single or dual vehicle. It would make more sense for self-driving busses to be touted for safety and the trolley problem instead (though, granted, they may be far less likely to be in accidents already, having professoinal drivers).

by rictic 8y ago

nkoren 8y ago

It's a more generalised problem than you probably think, and is a factor in virtually every multi-party accident. Even if the root cause of the accident is someone else's human error, this still leaves surrounding vehicles with a choice about how to respond to it: whether to take damage or inflict damage, and if the latter, then inflict damage upon whom. For an AV, that's a trolley problem. For a human, it's not. We don't rationalise about such things: it occurs in an eyeblink, and gets covered under "shit happens". We convince ourselves that the outcome is inevitable, just a function of physics. But it isn't. Unlike humans AVs will have the ability and obligation to make choices in such situations; this is critical for their harm-reduction capabilities. When all of those choices are to some degree bad (which is very, very often), then it's a trolley problem. Which is why people who work on AVs are legitimately concerned with them.

xapata 8y ago
Eh. Or you can just let the auto-driving auto-mobile take the same "shit happens" approach that humans do. Whatever the code happens to spit out, that's the direction it goes, ethics ignored.
- nkoren 8y ago
  Yep. All that stands in the way is every personal injury lawyer, class-action lawyer, consumer interest group, public safety regulator, competitor, and short-seller in the world. Shouldn't be a problem.
  - xapata 8y ago
    I'll believe that when I see it.

mmt 8y ago

I could be wrong, but my understanding is also that fatalities per event tends to be in the low single digits. There are just a high number of fatal collisions. That, of course, makes sense if cars generally have 1 or 2 occupants and accidents are usually single or dual vehicle. It would make more sense for self-driving busses to be touted for safety and the trolley problem instead (though, granted, they may be far less likely to be in accidents already, having professoinal drivers).