Resupply by AI. What could go wrong?
Army Futures Command’s newest Cross Functional Team-Contested Logistics head, Rob Watts, talked about resupply by AI and robotic delivery vehicles such as UAVs and roboboat.
“We want a tool to tell us, when we get to precision sustainment, that ‘you need to start moving some ammo to island X today because it’s going to take 10 days to get there, based on the threat information from the intel community,” Watts said.
Remember when MREs came out – Meals Ready to Eat – one of their nicknames was Three Lies in One. Kinda what I felt like reading this….so many things potentially wrong with this concept. The biggie, of course, is the usual: “based on the threat information from the intel community,”. Who wants to trust their asses to intel estimations? Of course you do, because they always get it right – ten days out – right? And who knows what will happen in those 10 days, eh?*
Even snapshot-type questions, such as how much fuel is available to units in three separate locations, should be answered in minutes by artificial intelligence, rather than having a soldier scour spreadsheets and radio disparate units, Watts said.
Tying all the data being collected through the various sensor systems the Army’s using could also improve planning.
One example Watts referenced was if a commander needed to extend an operation from five to 10 days. Such a system could answer that question accurately.
Assistant Secretary of the Army for Acquisition, Logistics and Technology Doug Bush told Defense News in October that the service has a high interest in a new watercraft that it received the first prototype of in 2023.
I would note this particular boat (pictured) not a roboboat. 100 foot long, about a 350 miles range. At least they didn’t use the term Littoral.
Category: "Your Tax Dollars At Work", Army, Marines
A better improvement to logistic needs would be 3D printers that can “print” metal parts.
That would reduce what you need to move to remote areas.
Need a left handed sprocket? No need to stock your best guess at what is most likely to break or have to order it special. Just pull up the STL file, hit print.
All that would be needed to move is the filament instead of tons of parts you may or may not need.
Edited.
AW1
That’s why back in the day there was a PLL and basic load lists. For those to whom may have fergetted PLL = Prescribed Load List. Thats the stuff needed by the end users of the equipment listed on the unit’s MTOE (remember that?); Now the basic load is that magic amount of Class1, 2/4, 3, 5 (you get the picture) stuff that the unit needs to do it’s thing (kill people and break $hit). If one was to return the basic principles and stop trying to reinvent the dam wheel we would all be much better off.
Rember as always, your milage may vary.
My understanding is that the Army has a bunch of ships. I watched (what I think I remember as) an LCU unloading something at a spot by the Intercoastal in Camp Lejeune in about 2002ish. One of the Army crew (Army crew!) told me the Army had a whole lot of its own amphibious shipping if not more than the USN in terms of ship to shore connectors. I had to look it up this morning to see if that wasn’t some false recollection from 22ish years ago. I found this:
https://en.wikipedia.org/wiki/List_of_ships_of_the_United_States_Army#Currently_active_ship_classes
Yeah, we (the Army) had a bunch of ships, to include Hovercraft. They trained the MOS their at Fort Belvoir, VA (or whatever it is called now). CENTCOM had a bunch. But we turned them in a year or two ago. Funny how now we need them back! Hey how about we throw in a few Close Support Attack Aircraft… you know A-10s!!!
Fort Harvey Milk?
Firebase Rupal.
Pear Harbor as well.
Ready Reserve Fleet.
Here are two of them that I remember quite well.
https://www.thedrive.com/news/39228/the-us-armys-largest-4×4-ever-the-gigantic-1000-hp-larc-lx-amphibious-vehicle
I can see AI messing this up royally. Sure, there’ll be extensive testing and fielding phases, amassing data to use for algorithmic analysis of units’ needs in simulated scenarios. The problem is that those simulations never quite match reality.
So, a computer program indicates that the 69th Armor Regiment needs X amount of fuel for operations in Ukraine. Testing was done at Fort Stewart over the previous year and the calculated number of gallons was exceeded by 20% as a safety margin. What the computer program doesn’t take into account is that the unit’s combat operations are fast-paced, sucking down much more fuel than the often ground-guided movements across Stewart, and that the operation kicks off in the dead of winter, meaning the Abrams and Bradleys are running nearly 24/7 just to keep the troops warm. Now, a Russian drone strike hits a fuel depot, and the projected resupply is still 10 days out because, well, AI estimated that they had fuel for another two weeks of operations.
AI’s tendency to make sh*t up it needs to sound “credible” will be our undoing here.
The two tours in Iraq that I lived off the FOB I kept in my fantasies force feeding the corrupt supply guy’s gen 1 MRES. It kept me warm at night.
MREs: Meals Rejected by Enemy.
Meals Rejected by Ethiopians.
Meals Rejected by Everyone.
My Rectum’s Engorged
The Army needs it’s own fleet? Really. And just how will that be manned when you can’t enlist enough troops to meet your ground pounding needs?
The army already has its own fleet.
https://en.wikipedia.org/wiki/MGen._Nathanael_Greene-class_tugboat
Yeah, I forgot about the Army watercraft. Not sure I would call that a fleet. Isn’t it mostly shallow water boats like tugs and landing craft?
Easily enough firepower to overwhelm Guyana.
AI? Sensors? What could go wrong?
Hacking. Incorrect programming (GIGO). EMP (solar flair or man made). Corrupted data.
Need I continue?
GIGO is the primary thing in my opinion. How is it going to know current supply levels and usage rates at remote locations? Sensors? Everywhere? How would that work? Low level switches on every storage rack?
You dump a dozen pallets of ammo into a FOB in the middle of nowhere. What kind of sensors are going to tell the AI when that FOB starts running low?
I don’t know how the Army does things, but I know in the Navy, reported numbers were regularly fudged to meet goals and performance indicators. We’d run out of needed consumables because Supply wouldn’t order more or then they’d be forced to admit that they’d pencil whipped the previous (several) reports to meet some arbitrary goal (often so the Supply Officer could claim that he’d saved X thousands of dollars on his fitness report).
When the numbers coming into the AI are imaginary, the numbers coming out of the AI are going to be equally imaginary.
Imaginary numbers. ((The square root of negative one))
When I was teaching Command and General Staff College I found that the logistics portion was the least appreciated by the students. I didn’t find it all that bad, just some basic elementary school level math that seemed to be the sticking point for the students. Heck they even allowed calculators in the class. That doesn’t mean I want a machine deciding when or where supplies should be sent in a combat zone.
Not just a machine; even worse it’s an “intelligent” machine.
AI and roboboats? Good luck with that.
(As if the enemy won’t intercept roboboats carrying supplies.)
The Good Idea Fairies who thought up this idea need to remember Murphy’s Laws of Combat Operations, such as “Things that must be shipped together as a set, aren’t.”
Relying on logistics based on intelligence estimates is obviously a bad fucking idea.
Stoopid People of The Week Thread dropping early? Or, are we adding a new Thread Category under the heading “The Good Idea Fairy Strikes Again”?
Yeah, if I’m the Troop on the front line, I really don’t want to have to depend on some guesstimating robot to tell me when and where I’m gonna need beans and boolitts. Naw, give me a Supply Daddy Claw Type, 1 each.
You want to solve the supply problems:
Does Ft Story VA still have their fleet of “boats”
Whatever happened to “just in time” or was that just a
civilian manufacturing thing? You know, like Raytheon,
General Dynamics etc.
Just in time only works for common items. If one needs an assembly that requires 6 months to build just in time doesn’t work; and yet so many people expect it.
Boeing… ran out of parts when Saudis dropped a mondo aircraft order and couldn’t do sh*t for 19 days of work. Bet those “expensive warehousing costs” saved by JIT (awful, according to advocates) weren’t so bad after that.
“Tip o’ the ol’ Hatlo Hat to Jeff LPH for the source material.”
Care to share?
AI is fashionable now. Every wannabee intellekshual and staff weenie wants to pad his resume by writing “thoughtful” articles. Perhaps I will write one explaining how they can be replaced by AI.
AI is fashionable because they see it as a simple answer to complex problems.
I work in an industry that provides technology solutions to “assist” a basically non-technological industry be more efficient.
I’m a technical trainer. When we sell a system to a customer, I train the customer’s people how to run and use our system.
It always amazes me what the end users of our products expect the system to be able to do. They want it to automate and be able to predict everything. I’m sorry, a computer program can’t read your mind and take action based on what you are thinking. It can only react to inputs it receives.
But that’s the draw of AI. It has “intelligence” right in its name, so it should be able to “predict” what we’re going to need and proactively do things before we even know they need to be done right?
Well…I suppose in theory. But no matter how artificially intelligent it is, it’s never going to be any better than the inputs it receives. The smartest thing (or person, for that matter) in the world is still going to give you bad outputs when you give it bad inputs.
Some “learning” algorithms, when provided proper feedback, can adjust their output to take into account past events and conditions, but they’re still just predicting things based on statistical analysis. It’s easy to say “you only need 120 rounds of ammunition a day” when you’re not the one being shot at…especially when you’re a computer program that doesn’t even have any concept of what being shot at is all about.
Proper logistics doesn’t plan for the average day…it plans for the outliers – the special events – the zombie apocalypse. Because some days just aren’t average and if you’re not prepared for that day…you die.
Yep.
“GIGO” is just as true now as when it was created back in the ’50s(?). Computers are stupid. All they can do is add ones and zeros. But they do it fast.
Science still doesn’t know exactly how the human brain works after decades of research, but somehow a few computer geeks are going to figure it out and create a written version in a year or two.
I was a 3043 (Supply Operations Specialist) in the Marines.
Our way of handling it was very simple.
You came in with a request for something, and we said “NO! You can’t have it. If we gave it to you then EVERYONE would want one; go away!”
Good news due to lowest bidder SkyNet will not arise.
Bad news it will still blow up the planet.