Autonomous AI military robot:-)

[Pawelk198604]

I wonder what would happen if small and medium-sized countries such as Poland or Ukraine, have invested in the development of weapons, I mean autonomously thinking drones and robots, I know that this is impossible in the present state of knowledge.

But maybe if small states were to have a better chance of defense against the aggression of the superpowers. and larger neighbors.

Had to give the gun thinking robots can to minimize the loss of life, which is particularly important if the state does not have large reserves of manpower.

[EvilotionCR2]

I would be more interested if small european countries like Slovakia or Austria started investing in it.

[Pawelk198604]

I would be more interested if small european countries like Slovakia or Austria started investing in it.

That would be something:-)

Investing in science is always a good choice.

[Kryten]

How would it help small nations? Superpowers would have overwhelming numbers of autonomous robots the same way they have overwhelming numbers of tanks and planes now.

[Tankbuster32]

By making them less dependent on manpower for military size. Anyway, I think a semi-autonomous system would be more optimal, so that the robots wouldn't just be able to fire at will, but instad be controlled and the targets be selected by a human. A robot doesn't have morals you know (unless programmed with them).

[Kryten]

By making them less dependent on manpower for military size.

Military effectiveness isn't proportional to manpower. It's mostly dependent on military development budget and the state of the arms industry.

[Coldfinger008]

No, just build small rockets with bomblet warheads. Really cheap ones.

Lots of rockets.

Then overkill everything.

Everything ._____.

[Mazon Del]

Autonomous lethal robots are quite possible in the current state of technology. However, autonomous lethal robots that have a low incidence of collateral damage is still a few years off. One of the big advantages of a robot though, is that you should be able to log everything it does and thinks about. This allows you to precisely determine why the robot decided to fire on what turned out to be a civilian. It is even a testable situation for humans. You could have a simulator that feeds in all the same data to a human that is 'controlling' the simulated robot and watch if the human decides to make the shot. Perform 30 or so simulations of this, ten with the target set up exactly as it had been, ten with a modified set of data that tended a little more towards a determination of a civilian, and ten with a modified set of data that tended a little more towards a determination of an enemy. If the difference in results between the 'real' simulation and the others is not statistically different, then the situation can be declared to have been ambiguous enough that it was a human-level judgement call. If the difference shows a statistical anomaly towards holding fire, then you know the robot made the wrong decision and you can now use this data to train the AI with a better set of constraints. Conversely, if the results show a statistical leaning towards firing, then you can prove that it made the same call as a human would have made (IE, even if the system had been a human-in-the-loop system, that civilian would still be dead) in which case you cannot blame the programming any more than you could blame the pilot.

This is quite important, because with a human pilot, they can be susceptible to faulty memory, outright lying, or even just the fact that the situation might have been somehow cloudy enough to skew their judgement in the moment.

[Omicron314]

South Korea has something along these lines, I think. It's called the Super aEgis II:

http://www.gizmag.com/korea-dodamm-super-aegis-autonomos-robot-gun-turret/17198/

[Sillychris]

A robot army is still reliant on manpower... you have to build the robots, maintain them, accumulate resources to build and maintain them. You have to transport them and tell them where to go. They presumably use ammunition and that takes manpower to produce. They use fuel of some sort and that takes manpower to produce. The list goes on.

A modern military's fighting force is in the miniority of its support force, and this trend seems to be continuing. There would be no exception for automated soldiers. The trend for automation of support forces should continue to a degree, but this ultimately just redirects where the manpower is coming from.

Of course, when all our industry is entirely automated and self replicating, the specific requirement of manpower can be eliminated... but in this case, you still need resources, energy, and robotpower. There are always limiting factors that are linked to the size of your economy.

The only way to stand against superior numbers and economies is to have better technology and tactics. This is a big part of the reason Germany was able to expand so quickly during WW2.

[Rakaydos]

With the costs associated with high tech guidance, often it's cheaper to use a "off the shelf" organic component.

In WW2, america expirimented with training pigions to fly bombs into targets. Pidgens being natually better pilots than the Japaneese Zero kamikazis of the same war. Even modern technoligy has problems stopping inteligent, self propelled, socially camoflauged terrorist suicide bombers.

Of course, the western world isnt big on throwing away people as part of consumable weapon systems. But the same concept applies to weapon platforms- any F22 could shoot down a dozen Predator drones with ease. And if theres one thing developing nations like india and china have, it's people- a self-sustaining resource.

[magnemoe]

First automated weapon systems is nothing new, mines and body traps are automatic and you had primitive versions hundreds of years ago.

Yes we all know dumb mines causes lots of collateral damage so they are mostly banned, however mines with an timeout or an enable/ disable function is legal.

During WW2 you got self aiming torpedoes, later self guided anti are missiles. Today we have automated air defense systems.

Common for all this systems is that you press the trigger and the weapon select an target in area.

An sentry gun will work like the automated air defense guns/ rockets like phalanx, you turn it on then you are about to be attacked and turn them off afterward.

Yes this has cause problem, think a jet was shot down by friendly fire during golf war 2.

You would only use the sentry guns on automatic during an major attack, against an small infiltration you would use them on remote control. You would however have them scan and report all the time.

And automated combat robot would mostly be an sentry gun who could move around.

[rkman]

But maybe if small states were to have a better chance of defense against the aggression of the superpowers.

Only if the superpowers do not invest in the same technology. But why would they?

[Tankbuster32]

Only thing is, that the people building the robots wouldn't be on the frontlines fighting, and as such would be safer. But you people do have a point.

Anyway another advantage of having semi-autonomous robot soldiers would be the ability to traverse dangerous areas such as minefields easier, as well as much increased accuracy due to an internal processor running ballistics calculations extremely quickly and highly precsie servomotors using that information. In short, robots make great snipers who can lie in wait even longer than humans, if you can solve the power supply problems.

[Seret]

Robots aren't going to entirely replace human troops any time soon, they'll complement them. The robots will provide air support, move materiel, do EOD/IEDD, run route clearance, stand stag and conduct close reconnaissance. The reduction in manpower requirements won't be massive, as mentioned above the majority of the manpower even in the army isn't frontline combatants. Robots will require logistic support just like people do, and they'll still require supervision from a chain of command just like people do.