Road to 350: What Does the US Navy Do Anyway?

[Lawman]

False dichotomy

What you get to do with these cheap point solutions, is very rapidly change those designs to specific missions or in response to enemy capabilities. Evolve these things inside the OODA loop of the opponent, while "quantity is its own quality"-ing them.


Indeed AI. Strap in, shipmate.

Automation is software loaded into controllers, and where I work, huge portions of software are now AI-generated. It is scary how well it works today, and that stuff is getting better daily at a scary fast rate.

Drone designs and similar are very often assemblies of pretty well-known tech into unique configurations that give unique capabilities. Guess what is really good at configuring?

Which is entirely dependent on a readily full shelf of inputs to manufacture and a combat environment devoid of impact to our homeland, something that will not exist in a near peer knockdown drag out fight.

Again, for all the “Drones and AI can do everything” magic, we just watched 5 million dollars worth of autonomous hardware get lost in the woods of a CTC rotation 4 months ago. Anduril and others standing their making promises followed by pikachu faces as the promise of what it could didn’t match what it did do.

Make a radio that doesn’t drop fills all the time or an airplane that doesn’t wake up stupid on its APU because it got rain water in the cannon plugs before we go promising the next generation of automated warfare. AI in the decision making process and automating staffs sure, but we are a long ass way from the Terminator promises of people like Palmer Luckey.

 

[BigRed389]

Which is entirely dependent on a readily full shelf of inputs to manufacture and a combat environment devoid of impact to our homeland, something that will not exist in a near peer knockdown drag out fight.

Again, for all the “Drones and AI can do everything” magic, we just watched 5 million dollars worth of autonomous hardware get lost in the woods of a CTC rotation 4 months ago. Anduril and others standing their making promises followed by pikachu faces as the promise of what it could didn’t match what it did do.

Make a radio that doesn’t drop fills all the time or an airplane that doesn’t wake up stupid on its APU because it got rain water in the cannon plugs before we go promising the next generation of automated warfare. AI in the decision making process and automating staffs sure, but we are a long ass way from the Terminator promises of people like Palmer Luckey.

I think both things can be true. AI is absolutely not at a point where we should have to worry about SkyNet taking control of fully automated Terminators any time soon.

That said, both AI capability and the computing hardware to enable it has been improving rapidly. By rapidly, in some areas, those measures have been improving exponentially. It’s not universally true, but across many key metrics, such as GPU throughput, and AI model benchmarks, it is.

Which is what makes it hard to predict when we’ll suddenly see those drone shows go from mere party tricks to deadly autonomous swarms, and the risk of falling behind that curve are significant.

 

[NoMoreMrNiceGuy]

Make a radio that doesn’t drop fills all the time or an airplane that doesn’t wake up stupid on its APU because it got rain water in the cannon plugs before we go promising the next generation of automated warfare. AI in the decision making process and automating staffs sure, but we are a long ass way from the Terminator promises of people like Palmer Luckey.

 

[taxi1]

Which is entirely dependent on a readily full shelf of inputs to manufacture and a combat environment devoid of impact to our homeland, something that will not exist in a near peer knockdown drag out fight.

Again, for all the “Drones and AI can do everything” magic, we just watched 5 million dollars worth of autonomous hardware get lost in the woods of a CTC rotation 4 months ago. Anduril and others standing their making promises followed by pikachu faces as the promise of what it could didn’t match what it did do.

Make a radio that doesn’t drop fills all the time or an airplane that doesn’t wake up stupid on its APU because it got rain water in the cannon plugs before we go promising the next generation of automated warfare. AI in the decision making process and automating staffs sure, but we are a long ass way from the Terminator promises of people like Palmer Luckey.

Full-on Terminator is a red herring. Cheap, rapidly adaptable tech with adaptation aided by AI is here right now and evolving fast. Seriously go back and read this report. This is all ongoing right now.

Ukraine’s Future Vision and Current Capabilities for Waging AI-Enabled Autonomous Warfare

As unmanned weapons systems reshape warfare in Ukraine, Kateryna Bondar explores advancements in AI-enabled autonomy. This report offers a comprehensive analysis of AI applications in the war and critically evaluates the extent to which human oversight remains vital.

www.csis.org

For example…

Ukrainian forces have widely adopted small and medium first-person-view (FPV) drones as platforms that may be quickly adapted for diverse missions through modular design and interchangeable equipment.
Not going to…have.

The Ukrainian defense industry is pursuing an approach of training small AI models on small datasets rather than developing large, all-encompassing models. This approach enables fast and efficient onboard processing on the limited computing power of small and inexpensive chips, which can be quickly updated, retrained, and upgraded to adapt to changing battlefield conditions.

Etc. It’s a good report.

 

[Random8145]

They have a robot using AI at the BJs stores now:

 

[Lawman]

Full-on Terminator is a red herring. Cheap, rapidly adaptable tech with adaptation aided by AI is here right now and evolving fast. Seriously go back and read this report. This is all ongoing right now.

Ukraine’s Future Vision and Current Capabilities for Waging AI-Enabled Autonomous Warfare

As unmanned weapons systems reshape warfare in Ukraine, Kateryna Bondar explores advancements in AI-enabled autonomy. This report offers a comprehensive analysis of AI applications in the war and critically evaluates the extent to which human oversight remains vital.

www.csis.org

For example…

Ukrainian forces have widely adopted small and medium first-person-view (FPV) drones as platforms that may be quickly adapted for diverse missions through modular design and interchangeable equipment.
Not going to…have.

The Ukrainian defense industry is pursuing an approach of training small AI models on small datasets rather than developing large, all-encompassing models. This approach enables fast and efficient onboard processing on the limited computing power of small and inexpensive chips, which can be quickly updated, retrained, and upgraded to adapt to changing battlefield conditions.

Etc. It’s a good report.

No don’t try and back peddle now.

2 pages ago you were having us marvel at a basic programming logic of drone dragon and dreaming of how amazing the automatous wave of death drones would be.

Now you’re pushing the ease of updates and adaptation.

These are not the same things. The second one has been around for decades, cheaper software cards and more plain English coding language models with the help of AI, just make it simpler and to a lesser extent less prone to isolation on an island in your own code language. Are we MOSA yet?

That isn’t the revolution in warfare that’s being sold by some, nor does it solve the problem of manufacturing in an input denied environment which Ukraine is not. Ukraine is effectively the world’s weapons laboratory right now where we make stuff in safe status and send it there to see how it works.

 

[taxi1]

peddle

Pedal

Oh, the angry autonomous swarms are definitely coming, I don’t think they will be Skynet.

Be patient. We are in the crawl stage of crawl walk run.

 

[Lawman]

Pedal

Oh, the angry autonomous swarms are definitely coming, I don’t think they will be Skynet.

Be patient. We are in the crawl stage of crawl walk run.

Oh did I make a Type professor? Sorry O was depending on AI to spellcheck that for me so I could do more. So what is revolutionary about an angry semi automated system that is technology dependent.

We’ve had area denial as a modern concept against operations since the invention of minefields and barrage balloons.

You aren’t revolutionary, you’re evolutionary at best in that model and while you can make the argument of concealment or total systems to cover an area that scales against the inputs necessary to go from a dumb mine to a smart drone.

A reminder to the room we’ve been in the “crawl phase” of HPM and DEWs for something like half a century. Those haven’t replaced anything yet effectively either.

The danger with your mindset isn’t that we fall behind, it’s that we spend all our energy and time chasing the next “WunderWaffe” while not paying attention to the fact that we can’t manufacture any of the munitions we are consuming to fight the actual war. Despite all its managed to prove with UAVs. And you have not in any of your posts once discussed the sustainment requirement on these systems or the inputs needed in their development and construction. When those current lines break down AI will help us realize we don’t live in a bubble, and supply chain dependence will force us into pick and chose moments like whether we want 20k dumb artillery shells or a handful of brilliant drones. The Ukrainians and Russians are consuming both right now at an exponential demand. There is a reason people like the CNO have been smart enough to say pump the breaks on quality and think about how we don’t need an adcap for every ship at sea and a cheap expendable will do. Same is true for Sams and surface to surface fires. The table at the trade show garnering a lot of attention is doing nothing to cut the price of a PAC-3 intercept by half or make them 6 times as fast as we currently can.

 

[Brett327]

I listen to a lot of the smart folks who really study the AI landscape… ideas about what is on the horizon, whether we’ll face any kind of serious alignment problems, and how we should think about regulation in that space. There are those who suggest that “something big” is coming, and it’s coming soon on the AGi front. I think it’s wise to temper those kinds of claims, given that a lot of these AI companies - big and small - are looking to cash in on this technology, and tend to over promise and under deliver, and we need to be careful of that kind of ambition when we’re devising policy, or pouring DoD resources into this stuff.

I have zero doubt that there’s great potential in AI to revolutionize aspects of our lives (for better or worse). I just hope we don’t step all over our collective dicks in getting there.

 

[Lawman]

I listen to a lot of the smart folks who really study the AI landscape… ideas about what is on the horizon, whether we’ll face any kind of serious alignment problems, and how we should think about regulation in that space. There are those who suggest that “something big” is coming, and it’s coming soon on the AGi front. I think it’s wise to temper those kinds of claims, given that a lot of these AI companies - big and small - are looking to cash in on this technology, and tend to over promise and under deliver, and we need to be careful of that kind of ambition when we’re devising policy, or pouring DoD resources into this stuff.

I have zero doubt that there’s great potential in AI to revolutionize aspects of our lives (for better or worse). I just hope we don’t step all over our collective dicks in getting there.

You mean like watching a bunch of brilliant idea guys talk about their new enhanced semi autonomous soldier borne reconnaissance capability, then prove they forget about basic things like the laws of thermodynamics and the effect of cold on battery performance.

They’re really good at proving how incredible things can perform at Dugway proving ground. None of them want to acknowledge what happened in Hohenfels at Combined Resolve. Somehow it being cold in Germany during January was an anomaly they couldn’t have accounted for.

 

[taxi1]

And you have not in any of your posts once discussed the sustainment requirement on these systems or the inputs needed in their development and construction.

Well, why didn’t you ask?

One of the beauties of things like sUAS is their simplicity and ease of manufacture. You don’t need a Lockmart or a McBoeing to make them, their manufacture can be distributed widely to 10-20 person shops or smaller. They can be built in a forward island chain. Huge expansion of supplier base without tasking the Raytheons and NGs.