Humans make very discrete and adaptive decisions based on multiple inputs from circumstance and context. An AI's decision making might end up being less adaptive and more general. How well an autonomous AI makes a decision will largely depend on how sophisticated its decision making and situational awareness capabilities are, and that is going to depend on human coders. Big problem here is whether human coders understand and can program the logic of these conditions well enough to not lead to unintended consequences. Maybe this isn't as challenging as we think because neural nets, but maybe we don't train our neural nets as well as we think we did because our training data informs certain patterns and characteristics we don't foresee.
Shenyang FC-31 / J-31 Fighter Demonstrator
- Thread starter RedSky
- Start date
- Status
vesicles
Colonel
Like any other weapons platform, AIs would be good for certain situation and bad for others. You don't use the F-22 as a heavy bomber. So eventually, we will figure out what kinds of missions are ideal for AIs, like defusing a bomb... We could develop a whole line of AIs specifically for bomb squads.
Another possibility would be crowd control. No human police officer will be thrown into an angry mob. Lots of times, officers are simply trying to protect themselves and make mistakes (like opening fire into the crowd). AIs and drones with absolutely strict rules of engagement will be sent in. They would be absolute prohibited to hurt human, even when they themselves are being damaged. In fact, these AIs won't even need to have arms and legs, just a block with some kind of wheel to move. Thus no chance of them hurting protesters. Like I said before, we simply cannot issue this kind of order to human police officers.
Yes, but that's a matter of how we choose to use our tools. It requires being sharply congnizant of their properties and limitations. Either way, if there is a failure it will always involve a problem with human judgement, somehow.
kwaigonegin
Colonel
Well when I said AI, I meant leaving the decision making skills to it as well.
As to the outcome, I agree that a computer with millions of computations per second will likely make less mistakes than a human however there is the psychological impact of such an engagement especially when life and death is at stake.
It's not too different than self driving cars. Would you let your car drive itself even if say some scientist shows you analysis and stats that says it's 10x safer than human drivers?
I wouldn't trust that mechanized krap as far as I could
No, 4 of those won't go in the bays, not gonna happen with the FC-31, or the F-35 for that matter, or PAK-FA or J-20?? NO
How did the discussion end up talking about drones and termiantors
I don't know, but I do know I wouldn't let these characters make a decision like that?? to assume that we would allow our machines to fight with other machines is naïve beyond belief, they will be used to "terrorize the op-for", so NO, its an insane idea!
SinoSoldier
Colonel
The length of the weapons (estimated to be between 4-5 meters) should be shorter than the length of the weapons bay (believed to be 1/3 the length of the aircraft). Each bomb weighs 500 kilograms.
you "might" get two in?? four, I just can't see that, they would have to be line abreast? and theres no way?
Here I'm gonna disagree wit ya bro, its a lot different than self driving cars,,, its an ethical question, and I'm gonna put the Ka-bosh on any such weapon,,, ISIS is bad enough with out and compassion, as kold blooded killers! NO
- Status