News for hackers. Stuff that matters.
Re:Deep learning pioneer Yoshua Bengio is worried about AI’s... by mda on Tuesday November 20, 2018 @02:52AM (#18492416)
Apparently they think about it yet execute the orders regardless. Just judging by the atrocities of all wars.
Re:Deep learning pioneer Yoshua Bengio is worried about AI’s... by TangoTrotFox on Tuesday November 20, 2018 @03:04AM (#18492459)
Do you think you shouldn't steal things from other people? Of course. I mean I think we'd agree that everybody, more or less, feels this way, right? Yet each year there are more than 8 million thefts per year in the USA alone. It seems that the latter fact there contradicts the former, but it doesn't. You can't compare things to a baseline of 0. You need to compare them to what things would be like if our proposition was true. For instance if most people believed that it was okay to steal things from people then there would not be 8 million thefts per year, there'd be billions. 8 million thefts is actually very strong evidence that most people believe you should not steal from other people.

And so too for atrocities in war. The non-zero number of atrocities in no way suggests that people just blindly follow orders, as a matter of fact. You need to compare these numbers to what would happen if people gave no consideration to morality whatsoever in their execution of orders. And what are those numbers? I'm not sure, but I expect vastly higher than the relatively few atrocities we can reference.

Re:Deep learning pioneer Yoshua Bengio is worried about AI’s... by rtpg on Tuesday November 20, 2018 @06:13AM (#18493036)
There's a more subtle argument, that the military as a whole views the word "atrocity" much differently than the rest of society. Though lots of society now agree with a sort of self-referential notion that our institutions and structures define what is an atrocity, so that whatever we do cannot possibly be an atrocity, if the rules are followed.

We see this sort of attitude when it comes to discussions of police brutality. Many police officers view the use of lethal force through the lens of self-defense in high stakes scenarios. Others view the use of lethal force through the lens of "police are not judge, jury, and executioner".

When people get shot as they run away from the police, even if each side of the debate might be rational in their own world-view, the conclusions are completely opposed.

Re:Deep learning pioneer Yoshua Bengio is worried about AI’s... by mda on Tuesday November 20, 2018 @03:15AM (#18492493)
I understand what you mean, but I am just pointing out that because of its nature no matter how hard people in the organisation think about morals they are the hand that do the atrocities, and these happen a bit too often considering. I have experience with military as well and the tone of "oh they think so hard about morality, consequences" argument just makes my blood boil.
Re:Deep learning pioneer Yoshua Bengio is worried about AI’s... by quaice on Tuesday November 20, 2018 @06:56AM (#18493178)
Then I would put forth that you don't have much experience with military leadership. In practice, morality, as well as cause and effect, are a constant discussion among military leadership. To add to this: It's the lower echelons of leadership that are more about numbers and efficiency and even lower is where it's about "getting it done".

The disconnect, I believe, is what one considers duty another may consider blind faith. There is a vast amount of room for ambiguity here.

Re:Deep learning pioneer Yoshua Bengio is worried about AI’s... by killjoywashere on Tuesday November 20, 2018 @04:46PM (#18498087)
Look at it from a machine learning perspective: moral hazard is the most poorly labeled dataset of all. Adding as much synthetic data to the mix as possible is the best you can do. We all acknowledge it's insufficient. But reading authoritative texts, writing as best you can, and most importantly, practicing are the only things you can do. What exactly do you want? No war? That's a political issue. No atrocities? As grandparent said, to err is human. Let he without sin cast the first stone.
Re:Deep learning pioneer Yoshua Bengio is worried about AI’s... by Mankrik on Tuesday November 20, 2018 @06:37AM (#18493109)
This appears to base the arguement comparing to the worst case scenario, we don't need genocidal numbers of casualities to determine if morality has gone out the window. And that's another thing; numbers. The obsession with numbers, body counts, injuries, all to justify the reason we are there. In vietnam the US was fighting guerrilla warfare, in the middle-east it's called terror.

None of this will be declassified for years, it's technically still not over yet, we'll be waiting a long time to know some of the things we still don't know now.

Re:Deep learning pioneer Yoshua Bengio is worried about AI’s... by all2 on Tuesday November 20, 2018 @10:06AM (#18494276)
War is, by its nature, atrocious. Any activity that places death as the solution to problems is atrocious.

There is a point at which we have to consider whether war is worth the atrocity. What is so important that entire civilizations would condone the slaughter of their 'enemies'?

I have some of my own opinions on the above question, I only ask because I think the question should be considered carefully.