So, am I the only one who finds this both intriguing and terrifying?
https://www.youtube.com/watch?v=q5qno5i1H3k
As a child of the 80's/90's, I remember thinking "the rise of the machines" was comic book stuff. Now, it seems, not so much. Don't get me wrong, I'm not in a panic or anything, but this is just a bit concerning. What sayeth the hive?
You should be afraid of the invention of the killbot, by which I mean an autonomous humanoid robot armed with lethal weaponry. Not because they might become self-aware and collectively decide to kill all humans, but more due to the inherent dangers of creating autonomous killer robots that humans will be able to control.
Here's a little day-after-tomorrow sci-fi to expand on the point:
The technology seen in that video has already come a lot closer to practicality since the video was first released, now that there are low-powered and compact neural compute units that could handle object-recognition-powered targeting in such a small package.
My wifi is named skynet, and my work computer hal.
Buy more ammo.
Always buy more ammo.
Robots running around, cars driving themselves, your iPhone X recognizes your face, predator drones launching hellfire missles far away from their operators... The killbots are right around the corner (if they aren't here already). As stated above, we should be far more concerned about their human directed actions and uses than the skynet scenario, and we should start worrying yesterday. The revolution will be automated.
I'm not sure any amount of ammo is going to be enough...
If you guys knew what has to happen before a hellfire is fired from an MQ-1 or MQ-9 you wouldn't be so worried.
I think the tech is evolving but there are more controls in place than you'd think.
In reply to The0retical :
Let's not let facts interfere with perfectly good paranoia. 
In reply to Dusterbd13 :
There's certainly a good reason to have conversations about uses of the technology and what level of automation is acceptable. There's always that one guy/nation state that'll take things too far so it makes sense to talk about that stuff now to get controls in place. I think there's too many adults in the room to allow the automation of killer machines however.
That being said I find things like the recent Google employee revolt about using machine learning to identify object using UAV camera footage absurd. The counter argument should be "So you're saying you don't want to make the operators more sure of what they're pulling the trigger on?" It's going to happen one way or another, helping identify objects and situations better reduces potential for mistakes.
Mistakes happen. At least with UAV's the mistakes are surgical/limited and there's a massive chain that needs to connect prior to a strike happening to prevent errors.
Also on the subject of AI. There's a lot of discussion happening right now about why AI makes the choices that it does. At the moment, as the algorithms are evolving, it's starting to become a bit of a black box much like the AI from Hitchhikers Guide to the Galaxy where there's an answer but it's opaque at best and nonsensical at the worst.
No one in the right mind should be trusting an answer generating black box.
In reply to Dusterbd13 :
I read that as your wife is named Skynet and nearly choked on my glass of water.
The0retical wrote:If you guys knew what has to happen before a hellfire is fired from an MQ-1 or MQ-9 you wouldn't be so worried.
I think the tech is evolving but there are more controls in place than you'd think.
I was mostly just interneting for sport, but I assure you I would expect and demand the controls should be very robust indeed. I would argue that what "has to happen", at the present time, to fire a missle from a US controlled drone has little to do with what might come of our mechanized utopia.
In reply to WilD :
I know, I'm just being "that guy."
I don't trust very many things as they apply to government (odd considering my previous career) but, at least for the next couple of decades, I trust most of the career decision makers who are overseeing this stuff not to go down the killbot path.
The0retical said:I think there's too many adults in the room to allow the automation of killer machines however.
I hope you're right, it's almost happened before and could happen soon:
https://thebulletin.org/semi-autonomous-and-their-own-killer-robots-plato%E2%80%99s-cave8199
https://www.ainonline.com/aviation-news/defense/2018-03-22/six-six-lrasm-service-entry-nears
I’m just waiting for some Johnny 5 antics out of Boston Dynamic.
AngryCorvair said:In reply to Dusterbd13 :
I read that as your wife is named Skynet and nearly choked on my glass of water.
Shame it wasn't a stronger drink! I hear gin & tonic is good for cleansing the nasal passages...
And I agree, always buy more ammo...
Boston Dynamics had a robot built like a kangaroo that could jump/ balance itself 20-25 years ago. Pretty sure I still have a VHS of the thing testing somewhere. If my memory is right, it had some long tether, but they've been working on this stuff for a long time. That either means that the tech is maturing very slowly, or they've got stuff far more advanced than this, and they're showing us this as a low-tech distraction.
The0retical said:If you guys knew what has to happen before a hellfire is fired from an MQ-1 or MQ-9 you wouldn't be so worried.
I think the tech is evolving but there are more controls in place than you'd think.
True, as there should be. But theirs a reason I turned down being an operator in the air force. There's that disconnect and detachment/distancd from who you're killing that's scary. And the fact ISIS operate with children on their back or shoulders to keep uav from bringing the rain. I'll leave it up to y'all to imagine if we do or don't anyway.
STM317 said:Boston Dynamics had a robot built like a kangaroo that could jump/ balance itself 20-25 years ago. Pretty sure I still have a VHS of the thing testing somewhere. If my memory is right, it had some long tether, but they've been working on this stuff for a long time. That either means that the tech is maturing very slowly, or they've got stuff far more advanced than this, and they're showing us this as a low-tech distraction.
It is maturing very slowly and has been limited by computing power. Dynamically balanced bipedal walking is the most computationally complex form of motion. Moving on four legs or hopping/shuffling on two legs is far easier.
Saw this today and thought of this discussion. Looks like lethal drones have moved down to the squad level. The article is pretty poorly written, but if I'm understanding it correctly, these are backpack portable lethal drones.
I assume the "mini-missiles" are an evolution of the XM25 airburst ammunition since Orbital ATK is involved.
It should be interesting to watch how that develops and if it ultimately makes it to deployment. I know at least with the platforms I worked on firing or not firing the weapon was a decision that came from pretty high up. It would be a big change to turn that over to the teams on the ground. It's not a Hellfire but those XM25 rounds aren't exactly screwing around either.
Most of the special ops units I spoke with, at least the ones we weren't flying overwatch for, carried around RQ-11 Ravens to do their forward surveillance, also an AeroVironment product. The biggest complaint is that they're not nearly as silent as they're made out to be, especially in an valley, even with the electric motor. There was a lot of blame placed on them a few times when the unit launched one then suddenly came under mortar fire.
Random side note: The writer of that article needs to be sent back to press release school. Even I'm not that bad.
You'll need to log in to post.
