Forum adverts like this one are shown to any user who is not logged in. Join us by filling out a tiny 3 field form and you will get your own, free, dakka user account which gives a good range of benefits to you: No adverts like this in the forums anymore. Times and dates in your local timezone. Full tracking of what you have read so you can skip to your first unread post, easily see what has changed since you last logged in, and easily see what is new at a glance. Email notifications for threads you want to watch closely. Being a part of the oldest wargaming community on the net. If you are already a member then feel free to login now.

 Overread wrote:


Heck chances are if you had a cardboard cutout of a person you could get a current AI to shoot it to bits thinking it was a person whilst a human wouldn't be fooled.
And even then change the angle of view a bit and the AI might get fooled again even if it learned that the cutout wasn't a person moments before .

There actually was a test of AI 'Sentry' systems that were intended to perform jobs on the level of the dullest tasks imaginable for human soldiers, like guarding areas etc.; actual soldiers were used to challenge these systems. They quickly got wise to the fact that AI was good at spotting humans that tried to be sneaky, but helpless against things outside its training parameters, and just sat themselves in literal cardboard boxes and crawled through the guarded perimeter like Solid Snake or Wiley Coyote

https://www.gamesradar.com/two-marines-fooled-a-military-ai-using-a-classic-metal-gear-solid-technique/

This is of course a funny write-up for the most part, but it illustrates the central point. Other anecdotes about e.g. self-driving cars getting trapped in circles of salt, or causing problems because they mistook traffic signs being transported on a truck as traffic signs that were valid for the road they were driving on exist as well. Cognition is hard, and it's amazing what humans do all day without really noticing it.

Glad I'm not the only one who immediately thought of the cardboard box thing

I do find it odd that they didn't add thermal imaging to the sentry bot though. But then they didn't just want a motion sensor, but a system that could filter out false positives.

The other thing is if it can't tell that a box moving on its own is a problem then the thermal imaging result isn't actually resolving that issue. It's patching over it with a rough fix.

Which might only work if the person inside the box moves in such a way as a person normally would so that the thermal image appears like a person crawling. However something as simple as them wearing some knee pads and lifting the back of their leg up might make the machine think the thermal image is that of a dog or other 4 legged animal and of no concern.


So you've still got the cardboard box moving toward the sentry position and still got the same issue of the machine not seeing that as a problem.

Tsagualsa wrote:

 Overread wrote:


Heck chances are if you had a cardboard cutout of a person you could get a current AI to shoot it to bits thinking it was a person whilst a human wouldn't be fooled.
And even then change the angle of view a bit and the AI might get fooled again even if it learned that the cutout wasn't a person moments before .

There actually was a test of AI 'Sentry' systems that were intended to perform jobs on the level of the dullest tasks imaginable for human soldiers, like guarding areas etc.; actual soldiers were used to challenge these systems. They quickly got wise to the fact that AI was good at spotting humans that tried to be sneaky, but helpless against things outside its training parameters, and just sat themselves in literal cardboard boxes and crawled through the guarded perimeter like Solid Snake or Wiley Coyote

https://www.gamesradar.com/two-marines-fooled-a-military-ai-using-a-classic-metal-gear-solid-technique/

This is of course a funny write-up for the most part, but it illustrates the central point. Other anecdotes about e.g. self-driving cars getting trapped in circles of salt, or causing problems because they mistook traffic signs being transported on a truck as traffic signs that were valid for the road they were driving on exist as well. Cognition is hard, and it's amazing what humans do all day without really noticing it.

I think it’s also us humans do so much stuff we just never really think about.

Consider breathing. It’s an autonomic thing. We never have to think about it because we just sort of…do it. And as we gain or lose fitness, we adapt our breathing to tasks with varying degrees of success. Balance is another thing we just sort of do - but can train to improve. And so on and so forth.

So when it comes to programming AI, it’s gonna be messy for a good while, as you do your program, give it a whirl and it does something unexpected and daft (like voting for Brexit) because your program and programmer(s) overlooked something. The closest approximation in my education was writing down instructions on how to brush your teeth. Sounds simple enough, but providing clear, accurate, correctly ordered stages is really bloody hard.

Humans can of course makes leaps of logic via what we call (well, science probably doesn’t, but I’m not a scientist) intuition. This can lead to great leaps forward in understanding, or help us solve a problem. It can also cause stupid results (like voting for Brexit).

 Flinty wrote:
It depends on the dataset you train the ai on. There are currently microphone arrays that can triangulate where shots are coming from.

If you're talking the one that's often used in American cities - ShotSpotter, I think, unless they rebranded again - it's also tech that simply doesn't work reliably enough to be worth it. See, for example, this report from the ACLU, as well as further sources cited within.

 Dysartes wrote:

 Flinty wrote:
It depends on the dataset you train the ai on. There are currently microphone arrays that can triangulate where shots are coming from.

If you're talking the one that's often used in American cities - ShotSpotter, I think, unless they rebranded again - it's also tech that simply doesn't work reliably enough to be worth it. See, for example, this report from the ACLU, as well as further sources cited within.

#

You'd be surprised how much Jerry Bruckheimer style 'safety' stuff has no to actual negative effect on actual safety - it's mostly a political problem, it's an industry that has a metric ton of ex-military, ex-police or ex-politician people in it as salesmen and consultants, and thus is rife with kickbacks and pork barrel politics. On top of that, safety is always popular, it's easy to style yourself as a 'councilman that does something' with safety products, and it's an issue that usually finds bipartisan support. The UK armed forced once let themselves get bamboozled into buying 'bomb detectors' that operated on the principle of dowsing rods of all things, to the tune of millions of pounds for what effectively was a handful of useless wires and a battery in a box. In the 2000s, no less, not sometime in the 60s.

 Mad Doc Grotsnik wrote:
Even just “quick peak” to take stock is something humans are instinctually good at. Not to say “split second and you know the layout of the room”, but with training enough to get some idea of where someone might be in concealment etc. Not to mention a robot or drone is dependant on cameras, which I can’t imagine can match a human for looking around and processing.

Give them radar, and sonar!

It turns out that many water companies in the UK still try to find water leaks with dowsing...
...meanwhile something like a third to a half of all treated UK water is lost through leaks before reaching users.

Dowsing seems to be a particularly persistent myth to dispel.

Even in the 90s a lot of TV shows had things like mind readers and dowsings appearing in a lot of regular TV shows. Even if they weren't super serious, they'd still present those ideas as being serious things that could be potentially true.

Couple that with the fact a lot of groups spent a lot of money on them and there's likely enough of a "well there's so much money and attention is must be true right" going on.




Then again cold-readers still sell themselves as mediums and we still have faith healers and heck even homeopathic medication (the UK even has a dedicated homeopathic hospital)



Granted I've noted that homeopathy responded to the constant critical assessment of it by splicing in a lot of herbal/home/natural remedies. Things that actually can work and do have science behind them. Enough that its confused people enough that the whole madness of "water memory" curing manages to somehow still work.

However like many things it often preys on the less well educated and the desperate.


So even in things that can cause real harm - fake medicine - we still have issues. Flat Earth is a growing group and things like Flat Earth and Dowsing are most times harmless to the average person. So they don't get the same push back as homeopathy.

 Overread wrote:

Then again cold-readers still sell themselves as mediums and we still have faith healers and heck even homeopathic medication (the UK even has a dedicated homeopathic hospital)

Granted I've noted that homeopathy responded to the constant critical assessment of it by splicing in a lot of herbal/home/natural remedies. Things that actually can work and do have science behind them. Enough that its confused people enough that the whole madness of "water memory" curing manages to somehow still work.

However like many things it often preys on the less well educated and the desperate.

So even in things that can cause real harm - fake medicine - we still have issues. Flat Earth is a growing group and things like Flat Earth and Dowsing are most times harmless to the average person. So they don't get the same push back as homeopathy.

It helps charlatans a great deal that famous and influential people are convinced homeopathy et al. work and are superior to real medicine, and promote them as alternative, for free:

https://www.the-scientist.com/news-opinion/the-unscientific-king-charles-iii-s-history-promoting-homeopathy-70544

The King is by no means the only proponent at that scale: Germany had a Bundespräsident that was very convinced of it too, shortly after the war, and founded an endowment that promotes 'homeopathic research' and such ever since; other celebs and politicians are similarly intertwined with homeopathy and other woo...

Getting back on topic, I think infantry (including robotic infantry) will always be the most fundamental arm of land warfare whilst society and infrastructure is built for humanity. Wars are fought over human stuff, and require holding human infrastructure, which requires human-sized infantry to control. Maybe only a genocidal purge wouldn't need infantry, and we have nukes for that.

This leads me onto my next point:

 queen_annes_revenge wrote:
I'm going to take this to a slightly different, philosophical level.

If we were to render warfare into something done completely remotely using machinery, with no people involved at all, no risk to any human life, then what's the point? Would warfare even happen? Underneath the geo politics and strategic and tactical lingo, when we really get down to brass tacks, Is warfare not simply a primal drive to destroy, subjugate or defeat a set of people who differ in some way?

It ties into the whole AI debate. Warfare is part of human nature. If we remove ourselves from it too much, it's no longer in our realm, so why bother? I may be wrong of course, but I think part of the reason people are still involved is because people need to be involved, on a spiritual level.

This is a fascinating take, and views war as a sort of.. blood sport? Violent pressure valve? It reminds me of the concept behind The Purge. Extremely bleak outlook on the nature of humanity.

Personally, I disagree. I think wars are fought for stuff. Resources, which can and usually does include people. Most wars are either from necessity ("our people are starving, lets nick their food") or greed for more power over resources. There are only really two military options to affect resources- denial or hold. Because capturing stuff is the goal, denial through total destruction (what non-infantry are generally capable of) generally impairs that and the only way to hold stuff built for humans is to stick humans in it. Replacing infantry with artillery doesn't let you hold the mine works or the dock or the apartment blocks, it only lets you flatten them to dust.

I think if wars were fought primarily for the violence of their action, they would be more brutal than they already are, and there would be more wars and less wars that ended very quickly with little bloodshed.

Concern there? One country is going to be First with that technology. And let’s face it, given budgets and all the brains and bits and gubbins and bobs needed to have a robotic legion? There aren’t many.

US, Russia, China, possibly the EU seem the most likely candidates.

At which point, you gain a significant advantage, as people are far less squeamish about loss of materiel than they are the deaths of the young men and women in our armies.

I genuinely dread that advent, because it’s a serious change in the balance of power. And given not all those with power are of a balanced mind? Well.

 Mad Doc Grotsnik wrote:
Concern there? One country is going to be First with that technology. And let’s face it, given budgets and all the brains and bits and gubbins and bobs needed to have a robotic legion? There aren’t many.

US, Russia, China, possibly the EU seem the most likely candidates.

At which point, you gain a significant advantage, as people are far less squeamish about loss of materiel than they are the deaths of the young men and women in our armies.

I genuinely dread that advent, because it’s a serious change in the balance of power. And given not all those with power are of a balanced mind? Well.

It's a fair point but I don't believe it removes the human component from the battlefield. We have so much technology available dedicated to stopping signals from cell phone jammers to Wi-Fi jammers and so much more. Once upon a time there was a saying about controlling the spectrum. This leads me to think there will have to be some form of local human support/command and control.
I've been considering how to defeat robots since Ice Pirates did it with a blanket. They will just be another tool and aside as assets likely targeted with airstrikes, or what have you, that would not be used against humans because of the rules and laws of land warfare. And if not attacked directly their support elements will be found and removed from the table in some form. They will have a foot print and a delivery method and all that which will be an issue. Humans do as well but I think it's a bit of a different thing to tackle humans vs machines. Your robots will have their uses for sure, the question is how to employ them. Clearing wire, moving supplies over weird terrain or into a known kill box where you don't want to send in people but in a larger capacity I can't see it. ( Not yet anyway.)
Just something to think over. Brains bigger than mine have already figured it out.
I feel like blinding the robots and turning them off is the strongest choice for fighting against them.

Tsagualsa wrote:

 Dysartes wrote:

 Flinty wrote:
It depends on the dataset you train the ai on. There are currently microphone arrays that can triangulate where shots are coming from.

If you're talking the one that's often used in American cities - ShotSpotter, I think, unless they rebranded again - it's also tech that simply doesn't work reliably enough to be worth it. See, for example, this report from the ACLU, as well as further sources cited within.

#

You'd be surprised how much Jerry Bruckheimer style 'safety' stuff has no to actual negative effect on actual safety - it's mostly a political problem, it's an industry that has a metric ton of ex-military, ex-police or ex-politician people in it as salesmen and consultants, and thus is rife with kickbacks and pork barrel politics. On top of that, safety is always popular, it's easy to style yourself as a 'councilman that does something' with safety products, and it's an issue that usually finds bipartisan support. The UK armed forced once let themselves get bamboozled into buying 'bomb detectors' that operated on the principle of dowsing rods of all things, to the tune of millions of pounds for what effectively was a handful of useless wires and a battery in a box. In the 2000s, no less, not sometime in the 60s.

He didn't just fool the British.

Tsagualsa wrote:
The UK armed forced once let themselves get bamboozled into buying 'bomb detectors' that operated on the principle of dowsing rods of all things, to the tune of millions of pounds for what effectively was a handful of useless wires and a battery in a box. In the 2000s, no less, not sometime in the 60s.

Not quite true.
The British government testing team declared them frauds, but despite that the Royal Engineers Exports Support Team was paid to promote the devices at a trade fair. The device was also backed by British embassies in Mexico City and Manila, through the then Department of Trade and Industry.

So we didn't buy them, didn't do anything with the info they were frauds and happily promoted them to others.

 Mad Doc Grotsnik wrote:
Concern there? One country is going to be First with that technology. And let’s face it, given budgets and all the brains and bits and gubbins and bobs needed to have a robotic legion? There aren’t many.

US, Russia, China, possibly the EU seem the most likely candidates.

At which point, you gain a significant advantage, as people are far less squeamish about loss of materiel than they are the deaths of the young men and women in our armies.

I genuinely dread that advent, because it’s a serious change in the balance of power. And given not all those with power are of a balanced mind? Well.

Some would argue it has all ready happened, and is over too.

Think of the Drone warfare during Bush II, Obama, and the Trump Presidency and we are now seeing near-peer and lesser peer competitors taking up the baton. Drone Warfare is no longer a US only proposition and has not been for around a decade now.

 queen_annes_revenge wrote:
I'm going to take this to a slightly different, philosophical level.

If we were to render warfare into something done completely remotely using machinery, with no people involved at all, no risk to any human life, then what's the point? Would warfare even happen? Underneath the geo politics and strategic and tactical lingo, when we really get down to brass tacks, Is warfare not simply a primal drive to destroy, subjugate or defeat a set of people who differ in some way?

It ties into the whole AI debate. Warfare is part of human nature. If we remove ourselves from it too much, it's no longer in our realm, so why bother? I may be wrong of course, but I think part of the reason people are still involved is because people need to be involved, on a spiritual level.

Warfare used to include a spiritual element to it. Status, ritual, religion. But not anymore. Warfare has always been motivated by a potential to gain. Wealth, land, pre-eminent attacking, political gain. And then there was the spiritual side, the warrior culture, the religion, the desire to be the apex culture, which fosters and enables warfare to be waged. But today, that stuff is obsolete. No amount of warrior culture or "spiritual" side is going to help you win a war of knowledge, science, and industry. The ongoing russo-ukrainian war is an example, the japanese empire was another one. Warfare still involves people because the conflict, and the world at large involves people. One day, we might evolve to such a technological level that we can isolate ourselves into a bubble and live the rest of our lives in want and in need of nothing. Labor would not be required, and everything and anything can be catered for by automation and machines. Or we won't. But if we do, then and only then would people or "infantry" not be required in war. But like you said, when we get to the point where we want for nothing, what then is the point of war?

Also, consider the idea of waging war using only drones and computers. We are simply not there technologically yet. People who run around saying humans are not needed because machines can replace them at their jobs do not understand the complexities of industry and economy. People are required to build those machines and factories, and program them. Its ridiculously hard work building and programming just an articulated robotic arm, and then you set it to do the task of simply cutting out square metal sheets. Now you want to create a whole factory of these, and have assembly lines and have them do more and more complex tasks such as cutting complex shapes, bending things, welding them, and assembling them? I mean, we're not even accounting for the fact that tools and tool-bits wear out, or many advanced machines have custom parts that are still not automated yet, and you can't just buy these spare parts from a big brand company. Its just as well that today you can buy most of what you need for assembly lines, and a factory at large from other companies, because assuming you even have the staff and knowledge required to build all these yourself will take you decades, and cost the better part of a billion dollars. And now you want to take the human element out of warfare and have war being waged entirely by machines, when we have trouble in peace time even replacing certain machine parts? If anyone ever took a long think over that, then they'd realize how ridiculous the idea of infantry being obsolete in war is.