Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The US Air Force successfully tested AI-controlled jet fighter

arstechnica.com

Post reply on HN
Print view View original

71–80 of 97 posts

Re: The US Air Force successfully tested AI-controlled jet fighter

#71

Post by LarryMullins »

hulitu wrote: post #39
ck2 wrote: post #15

It's not the fighters you have to worry about. It's the 24/7 always in the air drone bombers with AI that have zero hesitation to murder civilian targets on demand or by "accident".

You mean like the killings of BinLaden right hands, when a lot of civilians were killed as collateral victims ?

Avoiding that sort of collateral damage is why they've developed the AGM-114R9X, aka the "sword missile". It has no explosive warhead, instead it has 6 blades that unfold right before impact. It can turn anybody in one specific car into hamburger meat, while leaving bystanders on the side of the street unharmed.

These have been in use for a few years now. You can tell when they're used from pictures of the car wrecks they leave behind. A very distinct 6 pointed star shape cut through the car roof: https://www.bbc.com/news/world-middle-east-39109438

Re: The US Air Force successfully tested AI-controlled jet fighter

#72

Post by gopalv »

scottLobster wrote: post #48
josho wrote: post #12

What will be interesting is when we design a fighter that doesn't have to accommodate a person. Its size can likely shrink and its performance envelope can be completely re-thought. E.g. Similar to the move from pets to cattle in computing. A jet fighter can very much be considered disposable and swap safety features for additional payloads. The squadron could first coordinate in a pattern far more sophisticated than…

I'm thinking it'll be more of a hybrid model, where you'll have a human in a fighter jet with an accompanying drone swarm, where the fighter serves primarily as a missile truck/Command and Control system, with perhaps the potential for direct human intervention in certain scenarios. But I doubt pilots will be dog-fighting or doing their own bombing runs in the near future.

> I'm thinking it'll be more of a hybrid model, where you'll have a human in a fighter jet with an accompanying drone swarm

The "Loyal Wingman" project is already underway to build that, though not as a swarm.

https://en.wikipedia.org/wiki/Boeing_MQ-28_Ghost_Bat

India is doing the same thing with their CATS

https://en.wikipedia.org/wiki/HAL_Combat_Air_Teaming_System

Re: The US Air Force successfully tested AI-controlled jet fighter

#73

Post by CSMastermind »

zabzonk wrote: post #32

i am pretty sure we will see self-driving planes (hell, all commercial jets effectively do that) before we see self-driving cars.

I've been waiting for autonomous cargo ships for a while. There must be routes that are simple and direct that could easily be 90+% automated. I mean there's not exactly a lot you can crash into in the ocean.

Even if you need a human crew to handle the stuff around the coasts it seems like a good place to start with automation.

Re: The US Air Force successfully tested AI-controlled jet fighter

#74

Post by CDRdude »

alkonaut wrote: post #21
wing-_-nuts wrote: post #6

The military always claims they will have a human 'in the loop' to make kill decisions. That promise will go out the window just as soon as other nations arm ai enabled drones with anti air missiles. There's no time for 'human in the loop' when the first to act survives. Thus down the slippery slope we go. In 20 years time we'll probably have aircraft that do figure eights over enemy territory for hours before automa…

> That promise will go out the window just as soon as other nations arm ai enabled drones with anti air missiles. Ethics don't go out the window because of autonomy. It's dark, but it's not somehow easy to shrug off as "ok no human then no ethics". We already have semi-autonomous weapons such as CIWS (Goalkeeper/Phalanx) which are autonomous because it's very tedious for a human to sit idle and wait for threats and t…

There is a dark joke that I think is appropriate here:

Q: What’s the difference between a wedding and a group of insurgents?

A: I dunno man, I just fly the drone.

Re: The US Air Force successfully tested AI-controlled jet fighter

#75

Post by ceejayoz »

2OEH8eoCRo0 wrote: post #66
ceejayoz wrote: post #41

Earlier quoted context omitted.

"AI" controlled hardware doesn't care about anything . A human is more likely to say "wait, this order doesn't make sense" or "what if those are just civilians?"

You're missing my point. A robot can afford to wait and get it right because it has no concept of self preservation. The operator giving the kill/no-kill can afford to wait and get it right because they're not in immediate danger. How would a stressed human in mortal danger be preferable here?

> How would a stressed human in mortal danger be preferable here?

https://theintercept.com/2015/11/19/former-drone-operators-s...

"U.S. drone operators are inflicting heavy civilian casualties and have developed an institutional culture callous to the death of children and other innocents, four former operators said at a press briefing today in New York."

It's a lot easier to be callous towards pixels on a screen, IMO, let alone an AI model trained on past data that makes the shoot/don't shoot call. I recognize that combat stress can cause bad decisions to be made, but I don't think the cost/benefit calculus is as clear-cut as you're making it out to be.

Re: The US Air Force successfully tested AI-controlled jet fighter

#76

Post by hermannj314 »

The military industrial complex needs a pipeline of tens of millions of veterans and their families to continue to sway elections and policies in favor of growing the military complex.

If we diminish the need for humans in service, we dilute the viability of the pro-military contingent to continue to manipulate our democracy.

If removing humans from kill decisions reduces the size of our forces, it might be the feedback loop we need to finally kill the US military Leviathan.

Re: The US Air Force successfully tested AI-controlled jet fighter

#77

Post by ninjagoo »

So back in the day, NASA was doing some tests with using neural networks/AI to fly fighter jets. They used a modified F-15: NF-15B, Tail # 837 for this purpose - Intelligent Flight Control System (IFCS). [1][2][3][4]

There's a cool video of that jet following the lead jet using neural nets, but I can't seem to find that online.

With AI/NeuralNets having made significant advances since those days (2003?), presumably this test craft is a lot more capable.

[1] https://www.nasa.gov/centers/armstrong/news/FactSheets/FS-07... [2] https://arc.aiaa.org/doi/10.2514/6.2008-6985 [3] https://images.nasa.gov/details-EC03-0231-4 [4] https://www.amazon.com/Results-Intelligent-Adaptation-Simula...

Re: The US Air Force successfully tested AI-controlled jet fighter

#78

Post by rch »

hermannj314 wrote: post #76

The military industrial complex needs a pipeline of tens of millions of veterans and their families to continue to sway elections and policies in favor of growing the military complex. If we diminish the need for humans in service, we dilute the viability of the pro-military contingent to continue to manipulate our democracy. If removing humans from kill decisions reduces the size of our forces, it might be the feedb…

> removing humans from kill decisions

Hard pass.

Re: The US Air Force successfully tested AI-controlled jet fighter

#79

Post by scottLobster »

mig39 wrote: post #51
scottLobster wrote: post #44

Earlier quoted context omitted.

Yeah, but I think historically that's going to be an anomaly brought on by the unipolar moment of US/Western military dominance, not a moment of ethical enlightenment. The US and its allies were so dominant that they didn't really need landmines to win anymore, and landmines are one of the cheapest effective deterrents their potential enemies could use. So it became yet another requirement for acceptance into the "ru…

> The US and its allies were so dominant that they didn't really need landmines to win anymore, and landmines are one of the cheapest effective deterrents their potential enemies could use. If the US doesn't need landmines, why won't the US ban them? The United States is one of the nations that hasn't signed the Ottawa Treaty?

That statement should more accurately target "US allies". The US is at the top of the system and can largely do whatever it wants, no one's going to slap sanctions on us over landmines. However our (western allies) strength gives our allies the confidence they need to voluntarily give them up, and we can coerce other nations to give them up to be part of our club.

It's all a power play at the end of the day, the ethical argument is window dressing. Some people take window dressing very seriously, but don't mistake that for being the decisive factor in policy.

Re: The US Air Force successfully tested AI-controlled jet fighter

#80

Post by MonkeyMalarky »

hermannj314 wrote: post #76

The military industrial complex needs a pipeline of tens of millions of veterans and their families to continue to sway elections and policies in favor of growing the military complex. If we diminish the need for humans in service, we dilute the viability of the pro-military contingent to continue to manipulate our democracy. If removing humans from kill decisions reduces the size of our forces, it might be the feedb…

Aren't most people in the military in non-combat roles like support and logistics? You're talking about replacing the 0.001% or whatever that are fighter pilots with AI as if that would make a difference.
Post reply on HN