It does better than this:
DARPA, the US military research arm, has launched a program to train fighter jets to engage in aerial battle autonomously with the help of AI algorithms. The Air Combat Evolution (ACE) program seeks to create military planes that are capable of performing combat maneuvers for dogfighting without the help of human pilots. …
I remember a program (about 40 years ago) for cruise missiles (using PROLOG no less) designed to carry out missions autonomously. The program could dodge enemy fire, adjust for decreased maneuverability caused by damage, go for secondary targets if the primary target was unobtainable etc. Modern aircraft combat is a lot about energy management (height and velocity) and I would think that both strategic as well as tactical assistance would be appropriate.
Of course, it is sad that all of this technology is for killing people. An AI program for negotiating peace would be most welcome.
I suspect that was for the AGM-86A and AGM-86B air launched cruise missiles using the TERCOM
(Terrain Contour Matching system) which was VERY sophisticated for its day!
See wiki link:
At the time that missile was over a MILLION DOLLARS per copy in 1982 or about $2.7 million in today's US dollars!
These days I can do much higher end FULL edge detection and vector-based image recognition and categorization using cheap AMD and NVIDIA graphics cards at less than a $1000 U.S. per GPU card!
"And how many missiles have you implemented that in ?"
I did MUCH BETTER than just implementing my software in missiles!
My system does full 3D-XYZ SOBEL edge detection, FAST pixel-to-vector conversion, fully-automated REAl-TIME object hunting, recognition, categorization, tracking, targeting and fire-control from a database of 200,000+ objects including large ships, small boats, cars, trucks, bicycles, buildings, tanks, trains, houses, street signs, towers, dams, persons, animals, rocks, trees, terrain features, etc. etc.
AND the system can auto-track, target and do fire-control at a rate of 65,000 OBJECTS PER SECOND using just 4 to 12 interconnected (i.e. gridded) cheap and common $1500 AMD or NVIDIA GPU Cards !!!
...AND... I can do it at 10,000 frames per second (fps) using four multi-spectral cameras that are each 4096 x 2160 pixels resolution at 64-bit RGBA colour depth using multi-GPU card interleaving! Our detection and ranging sweep is 220 degrees side-to-side and 180 degrees up-and-down!
THEN I mate object recognition system to our multi-spectral imaging suite which is ALSO four x 4096 x2160 incoming pixels with STEERABLE individually-scanning optical photosites AND individual millimetre wave transceiver sites where EACH PIXEL is a scanning group consisting of a CMOS 32-bits Nyquist downsampled to 16-bits per RGB colour and greyscale optical photosite channel, a 16-bit integer depth/distance/transparency channel (Alpha), a 32-bit floating point current range channel that is MEMS-based STEERABLE RADAR emitter/receiver (500 MHz up to 80 GHz!) combination AND a 32-bit floating point LIDAR-based emitter/receiver channel ALL mated and converted to the RGB + Alpha + D1-Distance + D2-Distance multi-channel vision recognition system!
We mix all the channels together to allow for automated detection and ranging of objects that can fly as fast as 160,000 kph (100,000 mph!) in-atmosphere and out-of-atmosphere!
We ALSO mate the system to a pulsed coil linear induction system that uses interleaved ULTRA-FAST DISCHARGE TIME ultra high current capacitors (as fast as 1 millisecond discharge) attached to super-fast IGBT (Insulated Gate Bi-Polar Transistor) switches that allows us to interleave and super-pulse magneto-cooled rail gun coils that accelerate 60 cm up to to 2 metre long and heavy ceramic-coated tungsten and ferro-core rods to as fast as 160,000 kph (100,000 mph) at an interleaved firing rate as fast as 600 rounds per minute with a kinetic energy of UP TO 11 tons of explosive fire power for each salvo!
Basically, we can OBLITERATE ANY hypersonic Russian Moskvet or Chinese Sunburn shore-to-ship or ship-to-ship missile, ANIHILATE ANY 30,000 KPH North Korean MIRV'ed ICBM anytime before, at or after apogee and even hit incoming 75,000 kph meteors dead smack in the middle with our fully automated 10,000 fps detection and ranging systems!
YEAH! I think we've got missile and space defence in North America ALL COVERED NOW !!!
We have and ARE ready to deploy ANYTIME !!!
So what does YOUR missile defence system do again?
Of course, it is sad that all of this technology is for killing people. An AI program for negotiating peace would be most welcome. ..... swm
Be prepared for the AI Programming of Autonomous Virtual Machines Targeting, swm, IntelAIgently Designed to Search and Destroy SCADA Systems and Executive Officers/Accountable Administrations responsible for the killing of peoples ...... which is surely an inevitable natural progression most welcome for any negotiable peace in which AI Programming is a Prime Lead Component in Novel Elemental Change Projects and Revolutionary Evolving AIMissions.
Such is certainly something which quite rightly terrifies the politically incorrect and inept Status Quo which funds and orders kill attackers and acolyte defenders alike, for they have never been able to be directly targeted and personally identified before, and nowadays there be no place to hide where one cannot be found and billed/prosecuted and persecuted.
Indeed, as much as things change, much like the ways of today, albeit with different stars being highlighted and permanently removed from Future Eventual Programming.
Of course, it is sad that all of this technology is for killing people.
Not really. Once someone chooses to kill us, killing them first becomes morally justified, legally permissible, and frankly, a damn good thing all round.
If you seriously think you can negotiate with ISIS, Al Qaeda etc whose only goal in life is our death, then you need a reality check. You can't reason with unreasonable people. And someone who is convinced their crappy life can be exchanged for eternal life in paradise with 75 virgins and unlimited luxury, then you have nothing they want and nothing you say will change their mind.
The real question here, is why are we endangering the pilot at all? Better drone tech with remote flying will produce a superior outcome for our military personnel, who protect us while we sleep.
How about, and I know this is a loony pinko idea, not invading their countries and thus attracting their attention in the first place?
You say "you can't reason with unreasonable people." Quite right. People who have had their families and possessions blown to bits by "shock and awe" do tend to be a bit unreasonable. Perhaps, I don't know, not imposing the Sh
itah on Iran, not creating the Mujahideen as part of a proxy war with the USSR, not boosting Saddam to start a war with Iran after they so unkindly got rid of the Shah and then invading when he got the idea, first that he was going to get Kuwait as a reward for the war with Iran, and then that he might pursue a non-dollar oil policy....perhaps not doing those things would have meant no 9/11, no Iraq war?
The Arabs and Iranians might still eventually have knocked seven hells out of one another, but we might well not have been involved.
Hmm. ISIS may be bonkers but they are not totally unreasonable. They DID apologize to Israel after accidentally bombing them - you've got to wonder why haven't you? And they HAVE negotiated withdrawals. They are not entirely nuts. Everything is up for negotiation - for the right price - definite. Speak softly, carry a big stick (Teddy Roosevelt?)
As for drones. You hack the tech and turn the drones on the enemy that sent them (probably us). Same for all this top gun AI. if it's hacked it's game over...your whizz-bang super fighter flies INTO the path of danger rather than away from it. Or just nosedives like a certain modern airliner. It may even disable the ejector seat - just for giggles - or even activate it (probably want to airgap that one if not already done). Maybe, just to be clever, make the plane unflyable in a way that looks like a bug in the software - rather than a hack - as you'd want to keep your hacks secret.
People get carried away with tech as the possibilities are endless and it sounds so cool BUT there are always countermeasures.
If you seriously think you can negotiate with ISIS, Al Qaeda etc whose only goal in life is our death, then you need a reality check. You can't reason with unreasonable people. And someone who is convinced their crappy life can be exchanged for eternal life in paradise with 75 virgins and unlimited luxury, then you have nothing they want and nothing you say will change their mind. ... LucreLout
Surely that requires their own death rather than that of others. Can't see that as being a being/organisation to follow with their executive direction in the driving seat/on the bridge exercising Command and Control of Computer Communications with Stellar Bursts of Raw Core Information for Realisation of Novel Noble IntelAIgents.
That's a Hell of a Glitch to be stuck with.
I wonder if one can sue the Supplier re ... Sub-Prime Goods for Impossible Goals? :-)
I'll be amazed if that is thought not possible when it so clearly is. Although given the antics of the world, just another amazement to ponder on for Resolving the Solution with Questions to be Asked and Fully Answered to Fail Safe Standard Security Protocols.
Beware SnakeOil Sales ...... which fester and feast on the Scrap from Sharks. Their ACTive Defence Solutions are riddled with holes to exploit and service, server and experience.
"An AI program for negotiating peace would be most welcome."
1. The world has WAY too many evil people. When they can, such evil people take over a government (like Venezuela, Iraq, Iran, N. Korea) and do what they do BECAUSE THEY ARE EVIL.
2. You can NOT negotiate nor compromise with evil. You can only DEFEAT it or at least OBSTRUCT it.
3. The only way to ensure safety in the face of evil is STRENGTH. Hence, you have weaponry and tactics that are SO overwhelming, nobody screws with you.
I doubt AI could negotiate peace. At best, it would negotiate a cease fire, or even surrender. But peace? Once you appear weak to an EVIL DICTATOR, he's got EVERY reason and motivation to DESTROY you. You can't negotiate peace with THAT.
I'll stick with AI being able to KILL the OTHER guy fighting for HIS country, in order to keep MINE safe. Then, with any luck, we'll NEVER have to use it in war. Just having the thing will be GOOD ENOUGH.
Damn, you got there first.
Though in a military context, Boeing's (low) catastrophic failure rate would look pretty good. A handful of fatal crashes are just a price you pay for overall dominance.
I wonder who they'll source the electronics from if, for example, a Chinese vendor could supply technology that's ahead of anything they can produce themselves?
I guessing this is an intermediate step in air combat. Most of the current generation fighters can outperform the pilot. That is, g-forces and flesh bags are not a good mix. The AI isn't mature enough to other than simple things and air to air combat isn't that simple to do.
A chimpanzee in a paraglider could outperform most American fighter pilots according to the RAF people I've talked to.
Hell, W.E.Johns mentions they're shit as early as 1917, to paraphrase "don't cross the lines at less than 15,000 feet, so if you get hit you have some time to glide back" and the inbredistanis toddle over at 7,000 get their asses pasted by a circus - and Bigglesworth and company have to retrieve their derrieres from the hot cooking utensils.
But let's look at the choices for who programs it. It's like a Denis Nedry lookalike convention. Microsoft, Apple, Android script kiddies, gamers all hired and managed by Crapita, hell lets really push the side valve powered boat out - have IBM code it. You'd get a cross between Watson and Holly from Red Dwarf.
Imagine what it'd spout if you put it on a nuclear bomber? "Dead Dave, everyone is. Everyone is dead Dave. Yes, even Trump - what? A girl can dream.."
Seriously, what could possibly go right?
Reminds me of a scene in the Andromeda Strain where the character gets a chubby for the computer voice and then is brutally let down "She's 76 and lives in Oklahoma"
In comparison to whom precisely? There are various "experts" who agree that the Americans in 1917 in all areas had the self preservation instincts of a mole in a minefield. The only one who had any common sense was Pershing who said let's go to Berlin and that way they'll know they've lost. Shame no one listened to him. But maybe not - because most people alive today wouldn't exist.
Time after time they were advised to do things the way the French and English had learnt at great cost - and time after time they ignored the advice and got a pasting. The Canadians listened - were trained and deployed properly and their losses were much lower and they mastered combined arms operations in 6 months - 100 years later the average American still thinks that "combined operations" involve a combine, a field of barley and a Fordson Major diesel or Farmall rowcrop (it were good enough fer yer grandpappy so it's good enough to you).
And all you need to know is that after all that not a single lesson was learned for the next time around.. By anyone but the Germans ironically.
Pervitin for the win, literally.
That reminds me that before D-day (the next war, of course) my father did a combined ops course with the Canadians (Royal Navy/Canadian Army) and had something of a view from the stalls of how well it worked on the 6th of June. Juno was a success. The American sector, however, experienced considerable difficulties, partly because of very poor Navy/Army co-ordination.
Before 1917 the Americans had really been limited to colonial wars of aggression against much less capable adversaries. The 1940s lot didn't even have that. Your observation about farmers is fair comment.
I was just objecting to the use of Biggles and his author as evidence of anything military. I may be an extremely amateur military historian, but I like my books to have references and bibliographies.
This post has been deleted by a moderator
We're firmly on the road to introducing Yet Another Point Of Failure, congratulations.
I mean, we already don't know how the statistical analysis machines actually determine their results, but we do know that we can make them think an elephant is a banana, so yeah, let's plug many $100-million worth of hardware into the fuzz generator and expect military discipline and surgical results.
What could possibly go wrong ? Oh, I know : if they decide to plug the ejector seat into the AI as well. Oh well, Lockheed will be getting more orders, I guess, so success ?
How can they even come up with an idea that evasive maneuvers should be predictable ?
Anything predictable is the the easiest to attack and kill. It should be completely unpredictable and crazy. It should be a surprise to any opponent or foreign AI (which usuaully means Alotof Indians).
The algoritms should be the most secret in the world, so will usually be outsourced to a contractor with development offices in Ukraine, Russia, India or China.
well if you're gonna take a lesson from Anime, there are the gundams, and also Eureka 7, in which (for some models) pilots were pumped full of drugs to keep them at a peak during combat, after which they had to be pumped full of even MORE drugs to "bring them back down" so the other drugs wouldn't kill them.
science fiction in general has come up with some pretty good, and sometimes pretty frightening, ideas. In one episode of Stargate (the TV series), the gang went to a planet where remotely piloted drones were being used in combat (one chair controlled several of them at once), except that the enemy planes weren't drones, and the enemies were actually more like freedom fighters. That kinda put diplomacy on hold...
Is because former pilots are in charge of the Air Force, and won't accept unmanned fighters until they are handed their ass by enemy unmanned fighters. You can bet your ass China's military leaders aren't going to be so short sighted.
China will have unmanned fighters that will cost a tiny fraction of what ours do, because they will be disposable rather than needing all sorts of redundancy and safety systems to keep a human alive. They won't have to worry about G force limits humans can tolerate. They will be numerous and able to utilize kinetic kills to take out opposing fighters. The first time a squadron of US fighters are decimated by a force of enemy fighter drones 10x in number despite costing 10x less, former flyboy Air Force generals will be confronted by their failure of vision.
Humans aren't stupid; they'll learn from your (unlikely, frankly) scenario and adapt and win. That's why humans are still in the mix.
Yes suicide drones would be a PITA, but they're not unbeatable. Using decoys and sacrificial drones would be an easy and effective counter.
Voyna, Yanks did not have to counter. Kamikazes were militarily ineffectual. As it is, close in defense weapons are still improving. Thats why the Russians have hypersonic weapons and are developing more. The arms capability race continues. One hopes that none of these are used in hot war and reliable airbreathing rockets are developed that can be used for cheap space launches.
2 escort carriers, 1 fleet carrier, 44 other boats sunk, 300 damaged, and that despite US air superiority due to the Japanese lack of fuel - which was the main point of my post.
Had we not applied economic warfare, which was so successful in the Med., the kamikaze would have been far more effective. The Japanese were able to turn out a lot of planes, partly because they were unarmoured. Had they had the fuel, the massed waves would have been much bigger.
The Israelis are already seeing that Iron Dome can be penetrated simply by sending lots of rockets together.
perhaps not. Current rumours about the latest flying hardware investigations indicates reason is soaking into USAin military thinking, once the F35 disaster has ground its way to a much loathed past. Different platforms, including aircraft for different roles, whod a thunk ? Long terms plans for wetware flown aircraft are predicated on idea that humans manage unpredictable or novel situations faster and better than drones. eg startups with a rational idea versus corporates. I digress. In some situations like sub hunting or mine sweeping drones will do better. Doing a marauder attack requires much more variability to respond than mere silicon.
Sort of? As I understand it, it’s been rather a long time since there was a fight with anything close to evenly matched AirPower on both sides, so a lot of this is hypothetical.
When you have an Air Force and they don’t, you can afford to take more risks to get visual IDs, etc. When both sides are evenly matched, you’d want every edge you can get, which means starting to fire from as far away as possible. But since no one has actually needed to do it in so long, we probably don’t know exactly how effective it would be and what percentage of fights would devolve into close-range dogfights.
IIRC that was a mistake the US already made previously, in Korea? Sent the F4 Phantom in armed only with missiles in the belief they could destroy the enemy at stand-off range.
Technology has moved on considerably since then though. As mentioned above the limit is probably more political (IFF before firing) than physical.
"Are they still a thing in modern aerial warfare?"
The idea of superseding close-in aerial combat was thought up back in the 60's, probably by LBJ's whiz kids and others like them [who excelled at killing people unnecessarily by making stupid "smart" decisions like that]. It went so far as to take guns off of planes. This may have been talked about in the Top Gun movie from the 80's as I recall, or maybe I just saw a documentary about the REAL Top Gun [which used to be here in San Diego, I live only a few miles from Miramar, which is now a Marine Corps air base]. In any case, guns were removed from F4's but pilots insisted they be re-installed, as missiles were inadequate for close-in combat which was still happening.
And I don't expect this has changed at all.
Arthur C. Clarke's "Superiority" applies again.
Interesting. DARPA’s just articulated why I think committing to 1.5$T in spending on F35 is a huge risk on a 40-50 year time frame.
BTW, there are some indications F35 is maturing and unit cost dropping. It might not be an automatic failure at the tactical level for the next 5-20 years. It may even dominate, maybe. But spending so much leaves us with massive sunk costs once autonomous weapons tech matures.
BTW, there are some indications F35 is maturing and unit cost dropping. It might not be an automatic failure at the tactical level for the next 5-20 years. It may even dominate
Of the five kinds of stealth (radar, visual, infrared, sound, em emission) it is biased toward the first.
Due to compromises to provide V/STOL, carrier, and land-based versions with the same airframe design it is slow, expensive, needs a very powerful engine, and suffers from high wing loading.
It claims a large weapons load, but anything beyond what fits internally destroys stealth.
Due to the shape required for the lift fan, it is radar stealthy only from a limited forward arc.
Due to radar stealth being a 30 year old idea, people have come up with counters for it. It is really only stealthy over certain frequencies. Longer wavelength radars don't interact with the stealth shaping due to non-resolution of the shapes.
The fragility of the stealth coating means flying through rain is bad.
One report indicated that weight requirements for V/STOL operation mandated the removal of fire suppression equipment. Oops.
So far they are having real problems getting the gun to shoot where the sight points. And it is the wrong gun for a lot of applications. The size (25mm) makes it inferior for ground support to the 30mm gun on the A10, while providing about 1/7 the ammunition. It is the US favourite, a gatling cannon, which means it takes about half a second to get up to speed and thus full rate of fire, compared to the 1/20 of a second for the revolver cannon other fighters use (Typhoon, Rafale, Gripen), which is not good for the quick shots of air to air combat.
It needs a large manufacturer provided computer system to operate, which limits the number of bases... minimizing the targets for enemy strikes against aircraft and ground facilities.
While its powerful radar is touted as an advantage, the reliability of LPI (low probability of intercept) not giving away the craft's location, increases in computing power and broad spectrum SDRs makes actually using that radar riskier year by year. Ditto for extensive sensor networking among aircraft and other platforms.
The F35 does not have a built in IRST (infrared search and track) system, which have been continuously improving and now have a totally passive detection range often exceeding 150 km. The three previously mentioned aircraft all have this capability. It may be that the US manufacturer doesn't want prospective buyers to start thinking about IRST and that big, hot enigne in the back...
By rushing the production of F35s to preclude cancellation of the project, aircraft have not been as fully tested as with a full prototype, development, preproduction, production model. Some of the first aircraft have now reached the point where it is no longer economically possible to bring them up to current standards. In addition, it is now suspected that the planned airframe life of 8,000 hours for the F35B V/STOL model is currently about 2,000 hours.
Analysis of pitting the same dollar value of F35As against other fighters almost invariably results in the total or near total destruction of the F35s engaged. Remember that once the ball goes up, the F35s are committed, as they are slower than the other aircraft, and not stealthy while running away.
The F35 also has a per hour operating cost between 1.5 and 5 times that of the other fighters mentioned. This directly affects how many you can operate, and also retained pilot skill - a key factor in combat success.
At least so far, it has a much lower sortie rate. When operating from a secure base as a 'bomb truck', without significant enemy threats, the F35 managed about one sortie every 3 days per aircraft. If I remember correctly all the others could sustain a rate of more than 1 sortie per day per plane.
There's a lot more, but I am tired of typing... check it out yourself if you are curioius.
We may assume this is currently deployed and in use by some elite groups.
Because of the nature of research, the military always has the most cutting edge technology. When the first machine intelligences surpass humans and take over, it will certainly be military machines. Hmm, we really are doomed.
Biting the hand that feeds IT © 1998–2019