@Arcana hmm possibly. OpenAI used to have a robotics team but they disbanded it, and I know self driving cars arent going well at all despite >$100B thrown at them
it remains to be seen, I am inclined to think this won't be a big breakout year for robotics
@nyoom It's more that it's going to be the focus, might not see big breakthroughs, but it's what a monetary focus is on. Figure AI are a big focus of investment
I do feel like the investing into specifically humanoid stuff is often silly though. Feels like you're sacrificing some meaningful metrics just to have "whoaaaaaa... it look like person!?"
@nyoom The shape is fine, I've seen some excellent stuff from Boston Dynamics and their humanoid robot. I mean more the "humanoid robots" in terms of facially having human expressions and stuff
@Arcana@nyoom It's the opposite, those robots are mostly battery and a bullet to the battery will kill one a lot quicker than a bullet to center mass of a horse will
@scathach@nyoom Currently yes, I'm talking theoretically though. I imagine it would be possible to build an armored one and I expect battery technology to improve
@nyoom@scathach I do expect there to be advances in similar kinds of tech though. Not mounts necessarily, but robots with legs that are built for combat environments. With the recruitment crises going on at the moment, it makes a lot of sense and I've felt like robotic infantry and armor makes quite a lot of sense. Special forces and the officer class remaining human makes sense, but regular infantry can be automated soon I imagine, and would make sense. Especially with the kind of targeting precision such things can achieve.
@nyoom@scathach Soon = within the next few years so 2030 would be within that margin imo
I do expect it to start getting more and more focus with the whole "we have a recruitment crisis" and increasing tension
I don't think it's a coincidence that OpenAI changed their company constitution to allow military application, and are now investing into humanoid robots too
the DARPA self driving grand challenge was won in 2005, we still have very few self driving cars in the road
Atlas, the robot you linked, was debuted in 2013, it doesn't seem to have been deployed to anything noticeable
fully automated infantry seems significantly harder to me than Level 5 FSD, which seems quite far off---we can't even reliably get self driving cars to work when a human can remote control them
almost all the AI progress lately has been in learning to imitate human behavior in various mediums (text, art, code, etc.; older stuff like image recognition or sentiment analysis also fits in this). in areas where you can't do that progress seems much less imprssive (playing Atari is cool but seems much more primitive than what LLMs can do)
drones are mostly capable of replacing fighter jets today afaik, yet not only do fighter jets still exist, new ones are still being built
the B-52H bomber has been in use by the US military since 1955, the last one was built in 1962, and they are expected to keep flying until 2050. the US military is not eager to phase out components that currently work
afaik, humanoid robots currently cost >$1M to produce. it's quite plausible the typical light infantry soldier isn't worth >$1M. (Musk claims Optimus can get down to $20k, but he has a habit of overpromising so it remains to be seen imo)
in counter insugency operations, a significant part of what armies do is more akin to police work than traditional combat. humanoid robots seem like they'd be very bad at this kind of thing compared to humans
@nyoom@scathach Not discounting your observations, just my first thoughts on them
I think one part of this was too early, in that the gains they'd get in 2005 wouldn't fully achieve it from the provided money. Secondly there was less panic or rush in 2005 for such a thing imo
This does confuse me honestly. That video demonstration I sent is very impressive and basically a decade old now. I don't really understand why we haven't seen a big uptake in its usage, I could probably come up with a bunch of use cases for such a thing, I would how much they cost to purchase and operate?
Infantry does a lot of different things. While I think fully automating them is a harder problem, there's multiple things that can be done with them currently even that would be able to bolster current militaries. Given the potential for very precise aiming, I could certainly see them being deployed to shoot down the small scale suicide drones being deployed in places like Ukraine currently. I imagine as shipboard soldiers too they could be useful in the likes of the Red Sea conflict in shooting things down too. I also imagine that a robotic sniper wouldn't be too far away from being viable and is mostly slowed by the absolutely huge public outcry such a thing would likely cause.
A lot of the jobs infantry and soldiers do are kinda dull and repetitive, this is somewhere that automation could certainly come in. A bunch of the stuff my granddad did in the military could be AI assisted or AI conducted
Drones mostly operate in a different category, I don't think they're a direct competitor to fighter jets at all, drones are kinda their own new thing imo
Certainly, not wanting to do away with old tech is a thing which makes sense, but the second a major conflict goes hot, that stuff runs out fairly quickly imo
Yes, $1 million to produce is definitely too expensive and I imagine this is the main reason they aren't yet deployed. If they can be brought down to about 20k though, or even 100k, I imagine that they'll be far more likely to see action. The cost of them is too much now especially given the thing I'm going to mention in my conclusion
This is true yes, but with the recent rhetoric and considerations, I think a lot of militaries are preparing for and expecting a larger scale conflict fighting conventional forces again soon
Ultimately, I'd say I think the main reason they aren't used yet is mostly the high cost of production and running them atm in comparison to the relatively cheap infantry soldiers (who traditionally come from poor backgrounds and are considered pretty expendable), but also the fact that I don't think any country at this current time wants to be see as the country to deploy "terminators" basically. Especially given the moment that things like that do get deployed it's going to quickly become an arms race and I don't think anybody really wants that right now.
I would say this is just a sign that specific tests are relatively easy to "cheat" generally and the long tail of edge cases is usually really long
similarly, I'd guess that video demo is "rigged" to some degree. maybe they needed a hundred attempts to get it to do the obstacle course properly, maybe it was remote controlled by a human like playing a videogame, etc.
this seems like a decent point, especially if they don't have to do things like climb rough terrain. that said I'm not sure they'd be any more effective or cheaper than drones for such a task
seems a reasonable point, though getting enough data may still be tricky
I don't really understand what you mean here; why can't they just build a drone that's as large and capable as a fighter jet, if the AI for it is good enough?
yeah, a big war would burn through a lot of tech quite quickly, the interventions western governments have done against third world adversaries produces minimal losses though
agreed basically
I don't think the US is likely to fight another superpower (i.e. China) before 2030 and I think the likes of Iran aren't going to require a bigger conventional army than what the US currently has
It's very possible yes, it would be good to see more examples and methodologies
As above, but also, I'd say that as a remote controlled thing is still extremely impressive and would be useful in a combat situation.
"that said I'm not sure they'd be any more effective or cheaper than drones for such a task" not currently for sure which reflects the usage I suppose
"I don't really understand what you mean here; why can't they just build a drone that's as large and capable as a fighter jet, if the AI for it is good enough?" They're basically used as disposable remote control bombers. Fighter jets are currently used for checking potential incursions into air space and if you make a mistake with that you can start a war. They do have recon drones though. Fighting enemy aircraft just isn't very common in current wars so it's not something worth building for. Also remember that airforces tend to be much higher class. I think all pilots are commissioned officers in a lot of militaries. Drones are cheap and disposable, proper aircraft aren't considered to be. Plus if you want to take an enemy aircraft down with an automated drone, that's just a missile.
"yeah, a big war would burn through a lot of tech quite quickly, the interventions western governments have done against third world adversaries produces minimal losses though" Precisely, current wars just do not call for it. Accidentally massacring a village in Afghanistan due to a terminator glitch is a bigger PR disaster than the benefit you get from deploying a robot in the first place in such situations.
"I don't think the US is likely to fight another superpower (i.e. China) before 2030 and I think the likes of Iran aren't going to require a bigger conventional army than what the US currently has" I agree, but I think that's why they want to start working on it now. I think they 100% expect a conflict with China to come and the recruitment crisis scares them a lot I think, especially given the sheer manpower that China can bring
@Arcana@scathach I think the main takeaway is that a big reason this hasn't gone very far yet is there just isn't demand for it right now? it's only really useful for something like a war with China
@Arcana@scathach well drones are already useful, and I expect them to be fully automated soon to the extent they aren't already. we might well see things like self driving tanks soon too
I don't think the US is likely to run out of "tens of thousands of lukewarm iq grunts" in the immediate term to a degree that humanoid robots would be viable, especially when they can rely on third world proxies or mercenaries instead
I don't think right now there is much benefit in automating a full warship. I expect it to come possibly in submarines, but a lot of the US of warships currently is in gunboat diplomacy, and displays of power and readiness
@Arcana@nyoom The only weapons system that makes sense to be fully automated are anti-missile and anti-aircraft defenses like this since it's something people are uniquely bad at
@icedquinn@Arcana@nyoom No that's not true at all, brainwashing soldiers into being comfortable killing people is one thing nearly every military is very good at
@scathach@nyoom@icedquinn it’s the natural state to do it, civilians are a modern concept anyway because it’s not something that comes to people naturally
@Arcana@nyoom@icedquinn@scathach so anyway it still works because when you get to drafting civilians, your killing machine training isn't nearly as effective.
Part of the issue with My Lai is that the military did a cursory investigation and got yes-man Colin Powell to dismiss it, but wasn't successful. For the life of me I don't know why people don't remember this about Powell.
@Moon@nyoom@icedquinn@scathach Yeah, but soldiers aren't really trained to be "killing machines". Effective yes, and trained to be decisive in the face of the enemy which is important, but people acting like they're trained to killed civilians don't know what they're talking about. They're explicitly trained not to do that and conscripts are at a much higher risk of committing war crimes compared to professional soldiers.
People tend to forget that the US was deployed exclusively within South Vietnam, their ally. While there were certainly concerns and nervousness regarding some potential civilians given how the NLF operated, the US military were not trained to murder civilians, and doing so is extremely stupid in what is effectively your own territory.
@Arcana@nyoom@Moon@scathach
i remember watching some stuff about this and they referenced a book or two. idk right now.
basically supposed that viet nam was more about the CIA fucking around than anything else. a whole lot of psychological programs got green lit and tested back then because of it. one of the great hits was the "new standards men" where they basically just sent clinical retards because fuck it lets see what happens :neocat_magnify:
did anyone think it was a good idea? no. but they did it anyway. because, well, we can.
similar shit happened in iraq/afghanistan. there was no real reason for the USA to be there--it was just there to punish another country for trying to unseat the petrodollar system, play with the toys again, etc. part of why there was so much clusterfuck is because IT TURNS OUT that Mattis et all were just fucking around with new tactical theories and didn't bother to tell people they were being used as bait cars etc.
@icedquinn@Arcana@nyoom@scathach We were in Afghanistan because the Taliban was allowing foreign jihadists to train there as guests. We had been dealing with that diplomatically for decades and when Sept. 11 happened, we stopped being diplomatic.
@icedquinn@Arcana@nyoom@scathach Without getting into any unprovable assertions, I know we financed and armed jihadists there during Soviet occupation. But I am just saying, we did have a reason to go to Afghanistan if you go by the conventional explanation of events.
@Moon@Arcana@nyoom@scathach i think it was a video essay talking about the show generation kill and it being based on actual interviews with the soldiers.
one of the things we were doing was just fucking around with theories. sending people out on patrols/missions just to see what insurgents would do, without telling anyone that was the plan. and having no actual response for any of it.
basically, just fucking around.
a lot of the civilian casualties were blackwater iirc.
@icedquinn@Arcana@nyoom@scathach I know for a fact that we were trying out new military equipment and "modernized" techniques. So in a sense yes the war existed so that people could try those things out. For example in Afghanistan they tested out the microwave truck that could send out high-frequency waves that would make your skin feel like it was on fire "to disperse crowds". It was too brutal even for the military so they got repurposed and installed in US prisons to "stop riots".
@Moon@Arcana@nyoom@scathach sadly i don't think i have archives (just a couple [audio]books on the laptop from when i read over those) but there was a [unverifiable?] line of inquiry that the purpose of the united states military is to beta test the torment nexus.
they basicallly tell everyone its about protecting the empire but its actually just run by people who hate you, figuring out the bugs in next gen oppression systems, so they can roll them out later effectively.
it does make a lot more sense of an explanation than "apropos of nothing we blew up a country because it wasn't related to some bad pilots."
@icedquinn@Arcana@nyoom@scathach Iraq was also training for drone equipment. And I mean small surveillance and autonomous drones, not just predator drones. And not even just overseas. The military was using drones to spy on anti-war protesters domestically. People kept seeing a big "dragonfly". Well what do you know.
@Arcana@icedquinn@nyoom@scathach We also got a peek at the silent helicopters nobody still talks about. There's been silent, radar-invisible helicopters probably since the 1980s. One of them crashed while they were capturing bin Laden.
Add comment