Re:Pierre Sprey on the F-35 [video]
by wodenokoto on Wednesday June 18, 2014 @03:26AM
(#7908903)
We are pretty close to driving a car, we have beat human facial recognition, we have very good weapons aiming.
Tying it up to something with a very, very low false positive kill rate is the hard part, but we are very close with a lot of the components.
Those are things that are doable with pattern matching and big datasets. We're no closer to situational awareness than we ever were. Note we have self-driving cars that have precise databases describing the area they're driving through.
It doesn't have to have weapons control, even. If it's limited to an "AI scout" role it could just loiter around, find potential targets and flag them for a human operator. If the targets are reviewed to be valuable enough you could take manual control and fly an assault mission.
Things like this already exist in various forms and the U.S. government isn't too happy when our 'friends' try to sell some of them: http://www.washingtonpost.com/wp-dyn/content/article/2005/06...
The whole process of command and control, targeting, ROE, etc. is incredibly complex and far from easy. Drone's will do more and more, but I'm definitely not a fan of Hunter-Killers on the loose. Two of the fixed-wing losses during the opening of OIF were blue-on-blue Patriot shoot downs: http://en.wikipedia.org/wiki/List_of_aviation_shootdowns_and...
Now there's a terrifying prospect. Is this actually something we should expect to see any time soon? My understanding has been that, films aside, "AI" software is nowhere near the point where we might run a drone that can kill on it