Bang bang, you are dead. Recall that song? Look at how easy it will be to replace soldiers and their guns. This is a great topic to discuss. For the simple reason is you can use the small to kill the enemy. And be certain you got the mission done.
This is going to be a serious problem with LAWS. UN efforts to limit or regulate military AI may be failing before they even begin. Arms control advocates had reason for hope when scores of countries met at the United Nations in Geneva last week to discuss the future of lethal autonomous weapons systems, or LAWS. Unlike previous meetings, this one involved a Group of Governmental Experts, a big bump in diplomatic formality and consequence, and those experts had a mandate to better define lethal autonomy in weapons. But hopes for even a small first step toward restricting “killer robots” were dashed as the meeting unfolded. Russia announced that it would adhere to no international ban, moratorium or regulation on such weapons. Complicating the issue, the meeting was run in a way that made any meaningful progress toward defining (and thus eventually regulating) LAWS nearly impossible. Multiple attendees pointed out that that played directly toward Russia’s interests. According to the Russian Federation, the lack of working samples of such weapons systems remains the main problem in the discussion on LAWS…this can hardly be considered as an argument for taking preventive prohibitive or restrictive measures against LAWS being a by far more complex and wide class of weapons of which the current understanding of humankind is rather approximate,” it says and goes on to warn that too much effort to ban lethal robots could have an unintended chilling effect on AI generally. “The difficulty of making a clear distinction between civilian and military developments of autonomous systems based on the same technologies is still an essential obstacle in the discussion on LAWS. It is hardly acceptable for the work on LAWS to restrict the freedom to enjoy the benefits of autonomous technologies being the future of humankind.” More importantly, Gil approached the entire five-day meeting in such a way that any made any progress toward defining and thus, perhaps one day, regulating, killer robots very difficult, they said. Rather than look at serious proposals and position papers put forward by governmental delegations, Gil presided over a chaotic and ultimately inconsequential discussion of AI generally, barely touching on the stated purpose of the meeting during the five days. At one point, he even shut down ambassadors and delegates who tried to turn the meeting back to the work of defining lethal robots. “A lot states came prepared to talk about definitions. That’s what the mandate was” said one participant. For a governmental delegation “to put out a position paper like that, it has to get vetted through a lot of parts of your government… it was discouraging. It’s important that States feel like they’re vested in the process.” That didn’t happen, said the participant. But Bendett’s work also documents growing Russian interest in developing and fielding weapons that use increasingly sophisticated AI. In 2014, the Russian Ministry of Defense launched a comprehensive plan for the development of prospective military robotics through 2025. In 2016 the Russians launched an annual conference, “Roboticization of the Armed Forces Of the Russian Federation.” Bendett believes that Russian defense spending in AI will grow since the Ministry of Defense has at least 10 research centers looking at applications for autonomy in warfare. And of course Russian President Vladimir Putin has even said that the nation that leads in AI will rule the world. “Russia taking a defensive stance against an international body seeking to regulate weapons other than destructive nuclear bombs should not have been such a surprise. After all, in many international forums, Russia stresses the ‘sovereignty of nations free to pursue their own political/military/economic course’ as a cornerstone of an international order they envision as a better alternative to the unipolar world with the United States in the lead,” said Bendett.....snip~ http://www.defenseone.com/technolog...uilding-killer-robots/142734/?oref=d-topstory
Russia Says It Will Field a Robot Tank that Outperforms Humans..... The Russian military will field a new armed tank-like robot that “outperformed” manned platforms in recent exercises at the Alabino proving grounds outside Moscow. That’s what Col. Oleg Pomazuev told the Russian news site “Military Review” in late October. Pomazuev runs the Department of Innovation Research at the Russian military’s Main Directorate of Research Activities, or GUNID. The robot, called the Nerehta, can carry a 12.7mm or 7.62mm machine gun or an AG-30M grenade launcher. The Russian colonel did not say which manned platforms the Nerehta outperformed, but among his service’s vehicles of similar size and armament are the BRDM reconnaissance vehicle and GAZ Tigr. Russians are testing a wide variety of UGVs, from small IED-disposal robots up to large armed ones, said Samuel Bendett, an associate research analyst with the Center for Naval Analyses’ International Affairs Group......snip~ http://www.defenseone.com/technolog...nk-outperforms-humans/142376/?oref=d-dontmiss
The reason that Russia and likely the US would refuse to join in any such treaty is the impact it would have on systems that are already in place. Such systems already exist, in CIWS and their C-RAM counterparts, as well as in the AI systems that run our AEGIS class ships (as well as a more limited version in the PATRIOT and THAAD missile systems). Each of these has a fully autonomous mode where all firing decisions can be turned over to the computer. This has never been done in anything other than testing, but it is in place as an electronic version of "Final Protective Fire". In other words, the unit-equipment firing is about to be taken out or the crews are incapacitated so basically the computers make all firing decisions based on the greatest threat detected. As for killer robots, I really can not see that happening for a great many reasons. Friendly and non-combatant casualties being the most obvious, as well as simply they would be to easy to take out and then capture. We are not even really using much in the remote control capabilities for this kind of equipment, let alone AI variants.
Agree And because any such treaty is rubbish. There is no difference between pressing “Activate” button on m16 aimed at certain person, pressing “Activate” button on a Tomahawk programmed for a certain target and pressing “Activate” button on a Killer Bot pre-programmed to find and hit certain target. They will come that’s for certain. But not tomorrow. The flying killer bot in the movie is next generation of miniaturization, it might take another science revolution to make it. Such a bot requires a battery with like 100times better energy density. Remember ASIMO robot? 70% of its battery went into computing + processor cooling. Flying requires less computing power than walking, but still. Imagine how much power AI processor will eat… The cameras that we have are comparable to an eye of a dragonfly in terms of performance, but only the big ones. Micro cameras are not even close. No facial recognition in dimensions of 10 grams… So not tomorrow, not in next 20 years.
I have worked in computers since 1971 and agree fully with your assessment. Not only that the logistics of targeting 10’s of people individually would take some pretty hefty communication and computational power unless each bot were targeted to only one specific person. Still the limitations you speak of are there. Doable, but not with what we have now.
Im not so worried about killer robots so long as its still humans programming them. Sure, they could quite easily become a serious threat to us individually, but we can still choose collectively to shut them down once the problem saturates in public awareness. I worry about them self-programming over a network, like the one that Russia has designed its new 'Armata' tank to operate within. They clearly are expecting a 'Skynet' style battle-grid to be available in the near future, where all military assets are linked in to a virtual battlefeild and automatically directed by algorithms for tactical and logistical efficiency.
Depends on what you are talking about. Tomahawk is an autonomous killer robot running on its pre-programmed routine without external in-flight guidance. Extremely successful and deadly. If you are talking about replacing crew of Abrams MBT with autonomous system that can be given a command to eradicate everything that resembles a human or armor in certain radius in certain coordinates. Than forget about it. Such a system will be 100 times more expensive and 100 times less effective than a human crew. The good part is that the progress is advancing with extreme speed, new doors of opportunities are opening each year. The bad part: there is no concept in sight of how to make an AI that would have at least partial effectiveness of a trained tank crew. There is three semi sentient autonomous AI systems in Armata at the moment. The Driver, the Gunner and the tank Commander. Although many dispute if the tank Commander could really be considered Semi Sentient or if he is running on animal instincts . No other self-programmed systems. Everybody wants a total battle grid like in modern computer games (where every tank is constantly reporting its status, ammo load, intentions, to a HQ via a secured channel). Would be so nice. Not going to happen. Costs like hell and useful only against a technologically retarded enemy. Though individual or networked parts of such a battle grid is working right now and are effective and irreplaceable (I mean accountant office ).
I incredulously watched most of the video and I think it's insane. However, a lot of technology that advanced civilization has first been manufactured for military purposes.
Most do not even realize that the Tomahawk does not even use GPS for it's guidance. It still operates the same way it did in the 1980's. Inertial navigation, which then uses essentially photographs stored in memory to validate the location before making a change in direction. If there is a question, the inertial navigation overrules. They now have added GPS. In that situation, so long as both the photographic and GPS agree that it is off course, it will then overrule the inertial navigation. But the last 2 are only backups and checks, they do not do the actual navigation. The entire course is laid out and fed into it before it is launched, and it does everything from that moment on by itself. Humans are generally more effective in such things than machines. That is why in the M1 (unlike Soviet-Russian tank designs) we still use a human loader. A great many tests done over decades have shown that a human is simply more efficient than a machine in that task, less likely to make a mistake, and is less dangerous. In a Russian tank, if the gunner instructs the loading system to load a sabot round instead of an HE round, it is going to do just that. Then they will waste a shot and possibly give away their position by hitting another tank with a low effectiveness round. If the loader is human however, he will know the target (as designated by the Commander) is a tank and will verify the instruction (as well ass announcing what every round is as they load it). This gives the gunner a chance to correct himself before the round is fired. We already do have such a system in place, and have had it for over a decade. Blue Force Tracking (BFT) is a system that combines encrypted radios, GPS, and computers mounted in each vehicle to give operators and commanders a virtual space to conduct operations in. The system identifies each friendly unit in the area, and helps commanders control operations on the battlefield. It also tells other units where they are, to help prevent friendly fire incidents. If you see an APC moving through thick smoke and dust, you may not fire until it is to late because it might be friendly. But if scouts on the ground report it is a BMP-3 and not an M2 Bradley, and BFT shows no friendly M2 units in the area, the commander can authorize firing on it. But it controls nothing, it is simply another tool that can be used to coordinate the battle space.
Well no. If you are looking at cognitive process, than the human is more effective almost at every aspect, image recognition, planning, intelligence, tricks, decoys, e.t.c. In counting, or mechanical process, the machine will mostly outperform human in a narrow field (not at all times). Normal Autoloader in a Russian tank is comparable only to the best human loaders (talking about modern MBT with 120-125mm round). Normal, general issue human loader is unable to keep up, only the prize winners of tank competitions are comparable. Second important issue is that the autoloader performance does not change with fatigue, and fatigue is not an uncommon thing at war. Big issue – reloading on the run or moving on rough terrain. When the tank is jumping up and down like a wild horse human reloading is impossible. But the biggest win of the autoloader is that it requires much less volume to operate than a human. This means, that less internal volume has to be protected by armor. This means that you need less armor to cower the tank while maintaining the same thickness. This means that the tank is lighter. This means that you need smaller chassis and smaller engine to maintain same maneuverability. This means that you need less armor to protect the engine and chassis, this means that your tank is smaller and lighter. This means that it requires less fuel. This means that you need less armor to protect the fuel tanks. This means that your can make the engine and the chassis smaller. E.t.c. this is a cascade process. Total win of the autoloader construction versus human loader construction is roughly 12-8 tons. This is an enormous figure for a tank. With autoloader your tank can be 10 tons lighter, or it can have 10 tons more armor. But of course the autoloader has drawbacks. One less human to help maintain the tank. One less pair of eyes to maintain surveillance (This one is the most serious). Russian autoloader limits length of a AP round (and you want to have those longer and longer). All these issues can be resolved. French Leclerc is one of the most high tech tanks there is. Its autoloader does not have length limitation. But it is French. It is normal for Russian autoloader to suffer a malfunction for 4000 shots. When Leclerc was released, it was normal for French autoloader to suffer a malfunction for 40 shots. These are all technical problems. They can be and they will be solved. Cameras will give 360 view, Russian autoloader will be redesigned to fit longer AP rods as this is not a problem at all. French will perfect their machinery to run smoothly. The question is how long it will take . Everybody has such a system, it does not matter if it is on a, Laptop, Map or in the Head of a Roman Centurion. These systems range from poor to crazy to more or less. US system failed miserably in Iraq in the desert (problem with sorting out feedback from “friendlies”, when one considers a contact to be friendly, the other considers it to be enemy, the third one incorrectly describes its coordinates, and the tank is constantly firing on someone). Russian high position general was wounded, when he rushed to the frontline to direct his forces manually, when the Russian command system failed in Georgia (brand new system, excellent theoretical performance). These systems will be perfected, eventually, some day, may be. (The last sentence is like 4000 years old, or so) But these systems have little to do with a computer game or Star Wars film battle grids. The difference is on principal level. If you constantly see your Advanced Abrams MBT on the map, than you are receiving this data from somewhere, it is not God sent. This means that your tank is constantly transmitting its location, its heading, its speed, its damage level, its ammo to a friendly HQ on a secured radio channel. If this is the case. This means that your tank forces are advertising their presence to the entire world (even though the enemy cant decrypt their data, presence or absence of a radio signal is sufficient). This means that the enemy is already launching radio homing AT missiles, which will home on your tank. He has no need to develop complicated targeting system. He can shoot in to the sky. Your every tank itself is shining in radio frequency as a new year tree. And same applies to all your forces. Your every solider has to carry a radio beeper. Your HQ has to process millions of radio channels which is completely out of question, technically. And then suddenly, the enemy turns on a powerful wide specter jammer, or detonates a nuke. That’s basically it. You have lost control of ALL your forces . It is always the fog of war, forever!
The technologies being created by or for the military are beyond what people an imagine. But they also are raising serious ethical questions. I don't know if the video in the OP is just a theoretical fantasy, likely is, but the varieties of methods of killing people - 1 or millions - at no risk, no radiation, and if desired no material destruction - is reaching mindboggling and horrific potentials and capabilities.
I'm surprised that radio controls flying "toys" have not been used for a terrorist attack. A person can buy one online that can carry up to 50 pounds. The can be flown by preprogramming of visually by camera. A radio controlled large model helicopter that can carry many pounds would even cost less.
