[drhodes]

> because we don’t know how to program them to consider how their actions will affect other people.

We don't have to. For example, a ruthless train door could be made less ruthless, by having sensors that respond to someone who is nanoseconds late, to open the door, just that once, just like it's elevator door brethren, but only once.

The problem is that "inherently opportunistic people" will take advantage of the machine kindness and take an entire train full of people hostage.

The oppressive train door is a dictator everyone loves, maybe if they were also razor sharp, people would give them the respect they deserve. phwump. guillotrains. on time, every time.

[presentation]

But doing so isn’t the default state for computers. Someone needs to think about that, and put effort into building that. Making it so that some concerned citizen that wants to make the system less ruthless, also demands the instrumentation of the system to allow for that, cooperation between all parties to facilitate such changes - it doesn’t happen by default. So while it’s possible to make a system not behave ruthlessly with enough effort, it is default ruthless, for better or worse.