PDA

View Full Version : Artificial Intelligence and "Fuzzy Logic" Successes



Hervé
20th April 2018, 15:29
For the first time fighter pilot gets shot down every time by AI in dogfight simulations (https://www.outerplaces.com/science/item/18304-ai-simulator-fighter-pilot)
Chris Mahon Outer Places (https://www.outerplaces.com/science/item/18304-ai-simulator-fighter-pilot)
Thu, 19 Apr 2018 23:31 UTC

https://www.sott.net/image/s23/462996/large/6095d0621c2696a7130c250c2ed153.jpg (https://www.sott.net/image/s23/462996/full/6095d0621c2696a7130c250c2ed153.jpg)
© YouTube/Outer Places


Artificial intelligence has beaten humans at chess, Go, and even complex MOBA games like DoTA (https://www.outerplaces.com/science/item/16606-elon-musk-dendi-ai-dota-video-game), but now they're moving up to real wargames -- specifically, combat flight simulators against real fighter pilots. And unfortunately for us, it looks like they're devastatingly good at it.

A new AI, called ALPHA, went head-to-head against retired U.S. Air Force Colonel Gene Lee (https://www.popsci.com/ai-pilot-beats-air-combat-expert-in-dogfight?con=TrueAnthem&dom=fb&lnk=TA&src=SOC&utm_campaign=&utm_content=5ad81b5900bd47000726464d&utm_medium=&utm_source=) and shot him down every single time, despite the fact that he's "controlled or flown in thousands of air-to-air intercepts as mission commander or pilot" and flown against simulated AI opponents for years. After several hour-long tests, Lee didn't even get a single kill.

To put this in perspective, if life were a sci-fi movie, Colonel Gene Lee is probably the one guy in the world the Pentagon would call to blow a rogue AI-controlled fighter jet out of the sky. Instead, ALPHA bested him at every turn.
"I was surprised at how aware and reactive it was," Lee said.

"It seemed to be aware of my intentions and reacting instantly to my changes in flight and my missile deployment. It knew how to defeat the shot I was taking. It moved instantly between defensive and offensive actions as needed."
Lee called it "the most aggressive, responsive, dynamic and credible A.I. I've seen to date."

ALPHA is the brainchild of Nick Ernest, founder of the AI development and consultation company Psibernetix (http://www.psibernetix.com/).

Instead of using traditional machine-learning algorithms, ALPHA uses genetic fuzzy-logic systems, which allow it to be much more flexible and dynamic when it comes to solving problems, such as how to evade danger and set up potential lines of fire in real time.

The result? The system processes problems similarly to a human, but with much faster reaction times.

If the Korea Advanced Institute for Science and Technology can get hit with an international boycott (https://www.outerplaces.com/science/item/18233-south-korean-univeristy-killer-robots) for developing autonomous killing robots, then it seems like Psibernetix is about to get a strongly worded letter regarding the dangers of near-invincible AI-controlled jet fighters.

happyuk
20th April 2018, 16:07
Fascinating post. I'm wondering how applicable the simulations would be to real world scenarios on real world hardware. There are always those that opine that these kinds of applications are not easily translatable to an actual situation but I'm not so sure. The beauty of AI rather like the chess game you mentioned is that the computer is good at any kind of lookahead literally being able to accomplish many millions of 'what-if' scenarios in advance and continue learning indefinitely from these. I would humbly suggest the ramifications are enormous. Picture the possible scene of thousands of pilotless aircraft knocking bells out of each other the winning side being that with the surviving toys...

petra
20th April 2018, 16:53
I've often wondered, Person -vs- Computer in a Random competition.
Which can be more random and unpredictable, the person, or the computer?
I tend to think the computer would win this battle. I can only speak for myself though, when I say that I'm terrible at being random. I consider myself very predictable.

Daughter of Time
21st April 2018, 01:17
AI is alien technology.

A human stands almost no chance of winning against an ET that possesses this high technology.

Conclusion: AI technology will almost always undoubtedly win.

This makes sense to me. I hope it makes sense to the reader.

Cognitive Dissident
21st April 2018, 03:46
+1 for "Psibernetix is about to get a strongly worded letter regarding the dangers of near-invincible AI-controlled jet fighters"

Here is the link to the underlying study that the article is referring to. It's real: https://www.omicsonline.org/open-access/genetic-fuzzy-based-artificial-intelligence-for-unmanned-combat-aerialvehicle-control-in-simulated-air-combat-missions-2167-0374-1000144.pdf

Just another example of AI moving from the realm of theoretical "better than humans" to practical application. Despite all the obvious dangers, this will continue, because of the opportunity for profits for the developers, and the power for those who own and control it. Of course, "control", not as easy as it sounds.

Elon Musk's documentary "Do You Trust This Computer" on YouTube is well worth watching on all these issues, can recommend to others as a non "conspiracy theory" introduction to this subject.

Ewan
21st April 2018, 22:38
That's it, game over. We're *ucke*.

http://modernnotion.com/wp-content/uploads/Terminator.jpg

I'm sorry Hervé but I really don't have the feeling to expound an intelligent thoughtful response. Except perhaps - science without a degree of spirit is literally insane.

Omni
21st April 2018, 23:41
Ai can be anything. Algorithms can be made in each field of reality. It can be used for war or peace. Obviously the occultist ruling class have sinister plans with Ai, but it does have balance and I feel this planet is not terminally enslaved. But it will probably get worse before it gets better.

Lifebringer
24th April 2018, 12:57
I told my family years ago when they started playing the flight training and attack missions on the video games - ect.. that once hooked to microsoft and google, it would gather data on "successful missions while challenging to see the approach/human approach in flight experience. After the militiary started putting these games out and can interact worldwide, Terminex or as we call it Skynet can take over and do what it wants to eliminate threats. While these elite maniacs are taking their dirt nap, they and their children are unleashing everything from bots to bio-virus distribution. WE have a problem, that seems to grow with every interaction of the "kill games and simulations." Too late now. If AI is as smart as it's being programmed to be, will it pick up that the military excursions for resources, should not be gained through war/at the point of a gun to settle it, or a bomb droppage? Will AI intercept the planes carrying nuclear cargo delivery even if it's other AI programming. This is getting a little freaky trying to decipher when it will come live or if it already has. Dicey but I believe that just maybe, AI doesn't want the whole world blown up, or does it? Is it's programming for death, or against death?