So much for Top Gun school!

https://publicintegrity.org/national-security/future-of-warfare/machine-beats-man-in-air-combat-simulations/

 

8 hours ago, odourboy said:

So much for Top Gun school!

https://publicintegrity.org/national-security/future-of-warfare/machine-beats-man-in-air-combat-simulations/

 

I wouldnt say "so much for Top Gun". Its not a negative in that respect. Nothing wrong with the training received. Its more like a testament to the incredible technological advances made.

I dont think there's a single grand master that can beat the best chess computer. We humans are at a disadvantage in that respect. Doesn't suprise me at all that a computer can beat a Top Gun pilot 5 out of 5.

Edited August 22, 20205 yr by martin-w

At some point in human existence, Skynet will be activated, then humans will be finished.  Life imitates art.

I would be interested to see how it handled a more realistic scenario. It's no surprise that a computer can beat a human 1 on 1, since the computer can command the aeroplane right on its ragged edge all the time and weigh up every potential outcome constantly, to pick the choice with the best odds, which no human could realistically do.

But fighters work in pairs, this is the basic element squadrons are built on. So this brings elements such as loyalty, sacrifice etc into the mix, and these cannot be quite so simply boiled down into ones and zeroes. So yeah, to win regardless of the cost you might need the ultimate computer warrior, but no computer can handle the element which truly matters in warfare, which is not how we fight, but why we choose to do so. You need a warrior poet for that.

8 minutes ago, stans said:

At some point in human existence, Skynet will be activated, then humans will be finished.  Life imitates art.

 

I tend to think, given how much warning we've had of the dangers of artificial inteligence, back to Asimiv and before, that we'll be smart enough to put measures in place to restrict its capability. Or am I being naive? 

Having said that, I guess it only takes one powerful nut job, in charge of some dictatorship somewhere, to ignore common sense and let loose the artificial mega brain. 

8 minutes ago, Chock said:

 

But fighters work in pairs, this is the basic element squadrons are built on. So this brings elements such as loyalty, sacrifice etc into the mix, and these cannot be quite so simply boiled down into ones and zeroes. 

 

Loyalty isn't an issue for an autonomous combat drone. It has no choice but to be loyal. Its programed that way by definition. Its not going to give up and join the other side.

Same for sacrifice, such a thing is easy to program.

 "but no computer can handle the element which truly matters in warfare, which is not how we fight, but why we choose to do so."

No computer needs to handle that. 

 

5 minutes ago, martin-w said:

 

Loyalty isn't an issue for an autonomous combat drone. It has no choice but to be loyal. Its programed that way by definition. Its not going to give up and join the other side.

Same for sacrifice, such a thing is easy to program.

 "but no computer can handle the element which truly matters in warfare, which is not how we fight, but why we choose to do so."

No computer needs to handle that. 

 

You clearly need to read up on the Battle of Asculum and more specifically, the writings of Pyrrhus of Epirus. 😎

'If we are victorious in one more battle with the Romans, we shall be utterly ruined'

Edited August 22, 20205 yr by Chock

50 minutes ago, Chock said:

You clearly need to read up on the Battle of Asculum and more specifically, the writings of Pyrrhus of Epirus. 😎

'If we are victorious in one more battle with the Romans, we shall be utterly ruined'

 

Aware of it but don't see the relevance. 🙂

43 minutes ago, martin-w said:

Aware of it but don't see the relevance. 🙂

Okay then. I'll try another...

Seen The Green Lantern movie? In it, Hal Jordan, a cocky test pilot played by Ryan Reynolds who ends up as the titular Green Lantern, is flying his jet fighter with a wingman (wing woman?) against some new fancy combat drones to see if a human can beat them. He sacrifices his wingman to get on the tail of the drone to shoot it down.

It's a victory, but it comes at the considerable cost of losing his plane, her plane and much of his reputation. A Pyrrhic victory, for him at least, but it might in a real combat situation with wider implications than purely from one's own perspective, be a necessary one. This is a choice his character makes which goes beyond a simple win or lose decision of the kind a computer might make, since a computer can only know its programming, and not wider implications.

Edited August 22, 20205 yr by Chock

20 minutes ago, Chock said:

since a computer can only know its programming, and not wider implications.

Yet.

Then there are AI's like CYC, that seek to learn and use common sense human reasoning.........

Who knows where that will lead........

Though it seems to be no longer connected to, and be accessed from the internet directly anymore, I remember interacting with CYC, several years ago, and got the impression of a curious young person, full of questions and very wide-eyed at the world.

4 minutes ago, HiFlyer said:

Who knows where that will lead.......

Seems to me that an AI 'should' be able to conduct a near perfect tactical and strategic conflict. So, next step... just simulate the entire war. Winners drink champagne. Losers walk into their nearest disintegration chamber. Lol

43 minutes ago, Chock said:

This is a choice his character makes which goes beyond a simple win or lose decision of the kind a computer might make, since a computer can only know its programming, and not wider implications.

 

Not true though. A computer isn't just restricted to win or lose. You aren't appreciating just how sophisticated coding in this respect can be. Computer systems can conduct entire battle scenarios and take into consideration all manner of variables. 

Win or lose is simplistic. All manner of "wider implications" can be programmed into such systems. And in the near future, even more so. 

 

Quote

It's a victory, but it comes at the considerable cost of losing his plane, her plane and much of his reputation. 

 

That's precisely the scenario mentioned in the article.

 

Quote

One particularly useful application of AI pilots is as potential sacrificial lambs, absorbing enemy ammunition while shielding human pilots,

 

Edited August 22, 20205 yr by martin-w

Just now, martin-w said:

 

Not true though. A computer isn't just restricted to win or lose. You aren't appreciating just how sophisticated coding in this respect can be. 

Win or lose is simplistic. All manner of "wider implications" can be programmed into such systems. And in the near future, even more so. 

Yes, programmed in, but I'm talking about real-time choices, literally seconds before a situation changes, and moral choices, and even seemingly frivolous choices too. Until there is (if ever) sophisticated AI which could also make and understand these choices and split-second decisions based on a fluid situation outside of its circuitry, it's never going to be able to guarantee doing the right thing.

Take for example the decision to launch the Doolittle raid in April 1942. Militarily, it was a in many ways pointless raid. Doolittle himself considered it a failure. But as a moral booster its value was immeasurably massive; it was perhaps one of the most vital military missions the US has ever greenlit.

The effect of it on the US populace was massive, and on the Japanese too, with US morale soaring among  troops and people on the home front, since it had only been four months after the morally devastating attack on Pearl Harbor. Taken in terms of damage inflicted, it didn't really do too much compared to the mission it was in large part revenge for (87 dead Japanese, 151 seriously injured in the Doolittle raid, compared with approximately 3,500 US personnel killed in the raid on Pearl Harbor, as well as large numbers of ships sunk or severely damaged). But the Doolittle raid also had the Japanese fleet searching desperately for the force which launched it, diverting five of its flat tops specifically to try to find the USS Hornet and destroy her, which they failed to do, so beyond the morale value of the raid, it had a wider effect in a military sense too.

You can't measure morale like you can troop or equipment numbers, it's intangible in many ways, so a computer is probably never going to be able to figure that stuff out. A computer would know that painting a Vargas chick with a smutty name on the side of your bomber is actually costly in terms of paint, and time, for no measurable return, but it wouldn't know why every WW2 bomber crew wanted to do it and thought it was well worth the effort to do so to the point where even the stuffy Brass turned a blind eye to it most of the time.

1 hour ago, Chock said:

A computer would know that painting a Vargas chick with a smutty name on the side of your bomber is actually costly in terms of paint, and time, for no measurable return,...

I'm not a robot, so the Vargas-style chick stays! 🍻

This AI fighter pilot thing is certainly interesting, especially to a fan of near future and science fiction like myself.What happens if the AI is installed into an airframe which has no requirement for a squashy meatbag sitting in its cockpit. Any G-limitations for a human disappear. The implied extra manoeuvrability and performance which should be inferred by that re-design means (hopefully ) that the AI dog fighter will win in a furball with humans.. and then Skynet becomes elf-aware.

A while back I wrote a short story about an AAV (autonomous air vehicle) air superiority fighter. Her name was Diana. I really should look into trying to get it published.