SoylentNews is people
SoylentNews
posted by
cmn32480
on Monday October 31 2016, @06:16PM
from the explosions-killing-everybody-isn't-a-choice dept.
from the explosions-killing-everybody-isn't-a-choice dept.
Researchers at MIT have put together a pictorial survey http://moralmachine.mit.edu/ -- if the self-driving car loses its brakes, should it go straight or turn? Various scenarios are presented with either occupants or pedestrians dying, and there are a variety of peds in the road from strollers to thieves, even pets.
This AC found that I quickly began to develop my own simplistic criteria and the decisions got easier the further I went in the survey.
While the survey is very much idealized, it may have just enough complexity to give some useful results?
This discussion has been archived.
No new comments can be posted.
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
It's not reality that's important, but how you perceive things.
(Score: 0) by Anonymous Coward on Monday October 31 2016, @09:35PM
So I decided I'd pretend to be a particularly uninformed deontologist and prescribe that the car would always swerve to the other lane regardless because "you should always avoid your current mistake".
Their test came to several "interesting" conclusions about my morality: I 100% love youth and hate the elderly, I like women 50% more than men, and I always 100% prefer to save as many lives as possible.
Given my creedo, there's no way systemic bias could have factored into my choices, which tells me their sample sizes suck, and their conclusions are faulty.
I notice that you avoided mentioning that you 100% prefer to take action rather than not take action, which happens to be true and undermines the underlying message you are trying to convey that the survey is useless and MIT is stupid.
Putting that aside, though, you are correct that one person taking a 10 question survey is not an appropriate sample size. However, if thousands of people are taking it, possibly multiple times, then the sample size becomes much more credible. Moreover, any non-systematic biases should clear themselves out as more people do more surveys, because that's the nature of random numbers and statistics.
(Score: 2) by ikanreed on Tuesday November 01 2016, @04:41AM
I wasn't proclaiming it to be useless and stupid, just that with a relatively simple rule, they derived some completely incorrect conclusions about my thoughts.