US military takes aim at 2024 for human-versus-AI aircraft dogfights. Have we lost that loving feeling for Top Gun? The US Secretary of Defense Mark Esper said a “full-scale tactical aircraft” controlled by an AI system will go up against a human fighter pilot in a real-world dogfight in 2024. Although the experiment is a battle between brains and binary, it’s not aimed at replacing military personnel altogether: computers will be used for …
An attacking air force doesn't need to be good at dogfighting - just send in 1 more cheap drone than their fighters have missiles.
It'll be similar against ground defences.
The only defence is ensuring you have more missiles covering an area than they have drones.
I'm buying shares in missile manufacturers.
It was quite fascinating watching the AI vs Human dogfights. The scenario was limited to short range gunfights, and not sure if it included any damage modeling. Often the winner was heavily damaged in the initial encounter, so in reality whether they'd be able to keep flying, ie damage to control surfaces. Hopefully the next iteration would be more typical, ie a normal air superiority loadout & mission profile. I suspect the results would be the same. And then whether there'd be human vs AI drone, where the drone may have more offensive/defensive advantages as it doesn't have the meatware support systems onboard.
Bullets are cheaper than drones, we would just need a way to kill swarms of drones and the missiles can be kept for bigger fish.
For example, really good/AI aiming systems for turrets mounted rifles, shrapnel 'grenades', a big net, diversionary utilities c.f. chaff to shake drones off, outmaneouvering e.g. altitude etc etc
You won't eliminate dog fights, you'll just find other ways to fight before you get to that point, thats why soldiers still need hand to hand combat training.
Whether this will actually be possible in 3 years is another question
Your suggestions, but instead of flak exploding out of airburst shells, why not Kelvar netting to entangle in the prop/get sucked into the turbine.
Or how's about barium? An airburst shell that explodes barium all over the shop. Ten miles in and you have some relatively simple rockets knocking out every barium signature in the sky.
For example, really good/AI aiming systems for turrets mounted rifles, shrapnel 'grenades', a big net, diversionary utilities c.f. chaff to shake drones off, outmaneouvering e.g. altitude etc etc......Whether this will actually be possible in 3 years is another question
Been done, and in service, and apparently been effective in destroying drones, missiles and possibly an aircraft-
https://en.wikipedia.org/wiki/Pantsir_missile_system#Armament
And apparently can be linked into air defence networks (ie radar, fire control) so operators can pick an appropriate weapon system to engage targets. The 2x30mm autocannons can fire frag rounds for a relatively low cost way to deal with low flying drones.
Sadly something I don't think the West has an equivalent for, but drone aircraft would seem a good way to deal with enemy air defence networks like Pantsirs and other missile/radar systems.
> reams of text generated by OpenAI’s GPT-3 model.
I can just imagine some joker hooking up a text to speech box to this and connecting that to a call centre.
So instead of some plinkety-plonk music and occasional reminders of how important it is for customers to waste their lives on hold, we would be treated to computer generated speeches on the same topic.
Companies already do this. You call in and get dumped in a queue, then every 30 seconds they play a 45 second advertising spiel (yes, cutting off the end of it) after which they remind you that you're on hold and that your call is important, and recyle through the tail-cut ad again...
Although the experiment is a battle between brains and binary, it’s not aimed at replacing military personnel altogether:
:-) Oh please, you cannot be serious if binary is successfully overwhelming.
And it opens up yet another crack hacking front too for both before and after the fact, should it not be a fantastic fiction, for ...... well, would they be brainiacs?
And having had a moment or two to further think on that before posting, such a crack hacking front opens up even whenever it is a fantastic fiction.
I shall now rest for a while and ponder on such be within the quantum communication realm, where a this is also a that and something else altogether different and entangling.
In the Wild Wacky West, that is Spooky at a Distance, DARPA/IARPA Military Industrial Complex territory ......... although only by virtue of Private Proprietary Intellectual Property Holder Participation. ........ which on Earth invariably revolves around Titans of Stock Markets which do not prove themselves to be Phantom Unicorns and Wannabe Caesars.
> The European Laboratory for Learning and Intelligent Systems (ELLIS), set up to help Europe remain competitive in the field of AI and machine learning
That train left the station 25 years ago when Europe failed to produce a single global tech giant similar Google/Amazon/Baidu/Alibaba during the dotcom boom. Chickens coming home to roost...
The UK and the rest of Europe frown upon monopolies and will barge in uninvited to prevent any one company from taking it all.
In America, they chose to be more selective about who's wings they'll clip, and make encouraging noises towards many organisations as they suck up all the competition in their attempts to dominate the world market.
U.S. Marine Corps F-35Bs Are About To Go Aboard A British Carrier And They Might Be There To Stay
The UK government spent so much money building an aircraft carrier that it does not have enough money left to buy planes that they have to resort to "renting" out the space to the Americans.
This post has been deleted by its author
What does that sentence mean, if not, "not anywhere", or "banned outright"?
Why does legislation get written in a manner that a quick dash with a black marker can instantly erase half of the purpose, if not the entire reasoning behind it.
Would have a human piloted aircraft facing off against at least a dozen AI 'piloted' aircraft, since they will be so much cheaper to build - they could be built for far less if they are considered disposable as they don't have to protect a human pilot. Making them disposable also means they could add kinetic attack to their repertoire, something that hasn't been considered appropriate for human piloted aircraft since Japan in WW II.
In fact I wouldn't be surprised to see the cost differential be more like 50x or 100x, since they don't even need to design them to be capable of stealth or sustained supersonic flight. If you have enough of them you don't have to care if they can't keep up with a supersonic jet if there are more waiting 'down the road' so to speak. For kinetic kills, give them a solid rocket booster in the rear that can enable it to go supersonic for 5 to 10 seconds if it sees an opportunity to take down its adversary that way.
With any of these systems, AI or otherwise that either make the human remote or remove them from the equation altogether there is a huge change in the risk profile.
Because we do not generally support suicide missions in the West the human operative is important and expensive. If an aircraft is lost with the pilot surviving then the military will usually make attempts to retrieve them. This leads to the assessment of risk to then initial mission versus the importance of the target. Once you remove the human from the aircraft it becomes much easier as the risk is just financial. This is where the arms race comes in because it is now easier to take bigger risks. If the various weapons platforms are cheaper because they no longer have a human on board, commanders will be prepared to take much bigger risks against smaller targets because they can.
The more computers are then used to drive the selecting, targeting and delivery of ordnance the more distanced the human factor is. The Gulf Wars were bad enough with televised missile strikes etc this turns it into an actual game with no consequences to personnel involved.
The ultimate risk is that someone on the receiving end of this type of weaponry or who has this but is losing reaches a point where they have nothing to lose. This point is reached very much more quickly resulting in reactive decisions where time and impact is potentially measured in hours or even minutes.
You then have a very dangerous scenario where the country with all this tech driven remotely ends up being the target of what would be considered a terror device, nuclear or chemical/biological based. These are all available to the current list of disliked regimes most of which would consider the loss of a few of their lives as martyrs (or whatever) to be well worth the impact of taking out 100's of 1000's (millions) in the west.
The trouble is that those orchestrating the disaster will all be safe in their bunkers ready to throw everything they have got in the ultimate MAD.
Maybe I am wrong but the more we see how useless AI is and how stupid those in power are I am not that confident. There are far too many companies pedalling this shite to politicians and military brass who just lap it up. The Terminator Cyberdyne scenario may have been fiction and far-fetched but now some sort of fiasco based on that story-line is becoming possible.
I have said it before, humans are the smartest, most stupid and selfish animals on the planet and are ultimately going to implode and self-destruct. The only question is probably just how fast, big and permanent the collateral damage is.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
