Hacker News Reader

Comment by bhouston | Hacker News Reader

Why hasn't there been a focus on multi-role fighter drones? Is there a responsiveness problem or a contextual awareness issue?

I would think that most of the cost of this F-35 is dealing with the problems caused by putting human in the middle.

Having multirole drones would lower the cost a lot, and it would actually likely increase its theoretical effectiveness because you wouldn't have a pilot scared for their life in it.

It would be so scary to go up against a drone multirole fighter, it could pull off the craziest multi-G moves, and it would be fearless.

There must be some technological reason we have moved to this yet.

View on HN

Drones have a very real weakness: they rely on a radio signal.

But that aside, the F-35 program was never really about building a fighter jet. It was designed from the beginning as a way to funnel 100s of billions of dollars into the defense industry and to keep that money flowing for decades.

> F-35 program was never really about building a fighter jet.

This. It was primarily about international collaboration, not military capability (F-22 is a more capable fighter). Best military technology (with most capable pieces likely highly classified) can never be built as a multi-national project. And there are cheaper ways of boosting international goodwill than funneling hundreds of billions into favored defense contractors. Government :(. My 2c.

Why not drones instead of the F-35? They started the F-35 project in '92. And if we're gonna play "why not" with the F-35, why not 3 frames for 3 roles as all the aviation experts and real world economics suggest?

Drones of this capacity are coming in on the sides, are inevitable if the shit ever hits the fan, and provide tons of advantages. In addition to crazy manoeuvres and no issue being a kamakazi missile you've got machine learning enhancements and teams of remote pilots able to step in fresh to help. They also integrate well with existing forces.

In essence, though, some of the kinds of roles for the F-35 are roles we still want human eyeballs for the forseeable future. The drone model is untested as the default solution, so we "need" the aircraft anyways, and the end-product will be some proper SkyNet stuff people are likely to have problems with.

The tech is evolving, but the weapons program answer is that we need both solutions and the F-35 program is unkillable.

Open source autonomous flight:

https://github.com/andybarry/flight

The F-35 is already ancient, engineering data is likely compromised by potential adversaries and it mainly exists to create jobs in marginal electorates. If WWIII ever comes, the old ways will be forced to face modern reality.

So this is second-hand knowledge, but AFAIK current drones work well against technically inferior enemies, but foes such as, say, Russia, should be able to sufficiently disrupt communication to render them useless.

The Russians say their GPS spoofing is so good they can simply steer an incoming cruise missile back to its start point. What chance does a drone have?

That's a heck of a lot of ephemeris data to upload to a missile before it hits you.

Do we still need GPS for precision targeting? How far can good, recent, satellite imagery, sensors at a wide range of frequencies, and dead reckoning get us?

Also, could one build a local positioning system that has a few ground (or even airborne) positions broadcast their position and a time signal, with the missile using that info to compute its position?

> Also, could one build a local positioning system that has a few ground

In _Russia_ during _war with Russia_? Sure, go ahead, you drive the truck and I'll monitor your progress and provide snarky commentary from home.

>> There must be some technological reason we have moved to this yet.

The AI isn't good enough yet. We have self-driving cars running into things, fighter jets are an order of magnitude more complex.

You're also forgetting that most companies/people that are good at AI are refusing to do work that promote war.

I think I read a headline that some department of Google had the heebiejeebies about AI for the US military, but I think it's a bit broadbrush and also redundant to make the claim you have here.

China has huge AI potential, and I'm not sure you do or can speak for them; AI may increasingly be a 'package' that's brought into tech, so I'm not sure how much difference morals make in its development, if the end product has universal plug-in potential?

- ed

Unless, I suppose, the smart, utopianistic & pacifistic people who form AI in the first place deeply-embed moral stuctures that cause AI to somehow magically recognise its use for malintent and then auto-terminate... . Unlikely!

I bet the big defense contractors and three-letter agencies have pretty good AI departments.

Reply on news.ycombinator.com