The difference between the tech then and today are automated decision making capabilities. 20 years ago a turret could automatically target moving things. Now it can see humans, identify who they are, and decide who to kill without ever consulting a human. Basically, Skynet by next Tuesday.
Comment on Protesters Gather Outside OpenAI Headquarters after Policy Against Military Use is Quietly Removed
Buddahriffic@lemmy.world 9 months agoIt always was. There are no words anyone can say to prevent it from happening. That’s the unfortunate nature of arms races: if you boycott one, you lose it. With nukes, they involve things on a scale that can be detected easily, so nuclear nonproliferation has worked, to a degree anyways. But AI stuff isn’t detectable like that.
And I remember seeing a video of a high school kid who made an automated paintball turret around 20 years ago. We’ve had remotely controlled drones for longer than that. Autonomous drones are a thing already.
The technology already exists for that black mirror episode with the killer dog robots. It’s just a question of whether all of that has been put together yet (and I’d be very surprised if no one has done it), and today’s are probably easier to disable.
NikkiDimes@lemmy.world 9 months ago
Buddahriffic@lemmy.world 9 months ago
Yeah, all the advances in facial recognition and person tracking can be directly applied to drone targeting. Just need to handle aiming a camera and correlating the camera’s position with the weapons system. The only part that might be difficult is the processing power AI requires. But the camera feed could be streamed to another machine that sends instructions back to reduce those power requirements, but then the drone would be prone to jamming.
NikkiDimes@lemmy.world 9 months ago
Drones are already prone to targetted EMF guns, regardless of if they require wireless communication, so I don’t feel that be a significant issue.
Buddahriffic@lemmy.world 9 months ago
Until they become hardened against them. That energy could be absorbed into the case, reflected at random, reflected but targeted, used to charge the battery or weapons systems, or the circuitry designed in such a way that it doesn’t resonate and just passes through harmlessly. If a drone doesn’t need to receive an outside signal, it can be encased in a Faraday cage.
MIDItheKID@lemmy.world 9 months ago
“Now it can see humans, identify who they are, and decide who to kill without ever consulting a human.”
This is the technology that I am not confident in, and it makes it the most terrifying. Remember all the issues we have had with facial recognition not working very well on people of color? So instead of having cops misidentify POC and killing them, we will have robots that do it but faster and more efficiently. And if you thought nobody was held accountable before, I got some bad news for you.
TheDarkKnight@lemmy.world 9 months ago
Doesn’t China already have a killer dog prototype?
werefreeatlast@lemmy.world 9 months ago
Boss I don’t know why no one is buying our killer robot dogs!
How much are you selling them for? Here on Temu the prices are crazy! Still no one is buying! 29.99??? Wow!
28.99? Just give them away! C’mon people buy them! They’re almost free! Just come over and click the link below to Temu!
reverendsteveii@lemm.ee 9 months ago
I think you’re missing the part I was worried about, which is the part where military tech becomes police tech, and autonomous flying assassin robots are gonna be rolling down main street in a few years. They’ll say it’s to “protect our brave officers serving high risk warrants” but the police are already not responsible no matter who they kill and I don’t see that getting any better when they can just zoop a kamikaze drone in through a window and kill everyone in the house at once.
ipkpjersi@lemmy.ml 9 months ago
What a crazy dystopian future that will be.
Buddahriffic@lemmy.world 9 months ago
Which is also a good reason to make sure automated killbots are developed, because we’re entering a time where one person could decide to commit a genocide, press a button, and have a chance at seeing it happen. And the best defense against that is to already have friendly automated killbots that can react quickly quickly enough to deal with a killbot attack. Or to have other counter-measures. But even developing other counter-measures works best if you develop the target system along with them, otherwise you risk allowing your counter-measures to fall a step behind in the race.
All of this is inevitable. Avoiding an arms race is like a prisoners dilemma where everyone is better off if everyone cooperates, but any single individual (or group) can gain a huge advantage if they time a betrayal well.
reverendsteveii@lemm.ee 9 months ago
you’re proposing…what, private ownership of automated killbots to counteract police abuse of automated killbots?
Buddahriffic@lemmy.world 9 months ago
I think the main thing I’m proposing is that the future is looking pretty bleak in some ways and that trying to avoid that outcome might instead cause it to be worse.
That is a bit of a non-answer though. I think the best way to handle it would be like the 2nd amendment should be handled: that well-organized militia bit that the supreme court for whatever reason decided isn’t actually important. That could still get messy, but the state monopoly on violence is already pretty messy and is essentially just a ruling class monopoly on violence.
Give too many access to that power and random violence increases. Give too few and you risk getting fucked if the wrong people end up in charge of it. Finding a compromise between the two could still result in half of them deciding to go to war against the other half or something like that.
Ultimately, I don’t think there’s a perfect solution; it’s the same problem as trying to achieve world peace as a species that is capable of murderous rage and murderous cold intent.