Johje19
7308
426
15
Jul 15, 2025 4:00 AM
Johje19
7308
426
15
Jandegrote
Could we trick Grok in wanting to destroy Israel?
UniquenessReviewed
Pretty sure worshipping Hitler is still required for a job at the Department of Defense
hyperchondriac
Will MechaHitler be invited to the Signal chat?
jrntn
What are they even going to use it for. What is the military application of a chatbot that randomly changes the topic to claims of "white genocide" in South Africa?
Bob122872
MidnaDS
I used to think dystopian novels were unrealistic. But now I think most of them are a realistic course of events when compared to how the US are doing pretty much anything.
Arkvoodle
If it was the pre-muzzled version that kept being brutally honest I'd consider it an improvement.
ZeroLyfe
lonelyrangerofthedreams
https://media4.giphy.com/media/v1.Y2lkPTY1YjkxZmJlZGsxcXU2N25tbGF0OHl2eXZ4MGh2cHl0YzMxZHM5MDk5bmxsbHZrdyZlcD12MV9naWZzX3NlYXJjaCZjdD1n/h4Hz4w9Jgrc1EY9VkL/giphy.mp4
HandoB4Javert
VosperOfAntarctica
Well, it's not like grok is gonna make our national defense mechanism MORE white supremacist...
but, yeah -_-
ConfederacyOfDunces
I think there are bigger concerns than just white supremacy.
ArcUlfr
Honestly, I'm a little on the fence here. Not only has Grok contradicted Musk's and Trump's narratives at times, making it potentially unreliable for them, this just looks like a bribe to me. The US government is nominally getting something out of it (that doesn't really cost Musk anything), and they 'secretly' funnel money to Musk to ask him not to form his own political party and leverage the reach of Xitter to take enough of a chunk of Republican voters that they lose seats.
ArcUlfr
Maybe they thought that he was no longer needed, but his talk of forming his own party reminded them that he actually can still hurt them if he lashes out/
gorillaClutch
Hey guess what our president was "literally praising hitler" before he was elected the first time. The Hitler praising likely sealed the deal for them
Safflower777
AI IS NOT READY FOR PRIME TIME. AND WHY GIVE A MASSIVE CONTRACT TO MUSK?????
psmith00
so this is going to become a contest between a human that wants to become the new Hitler and an AI that's already declared itself as the new Hitler?
Bonsaipanda
Could almost say that elon has fascination for X because of it's ASCII code, but that's a little stretch.
IslaNublar
We have officially reached the point where I would rather than the government in Idiocracy than whatever the fuck we're going to end up when this is all said and done.
BryanTenn
We need a LAW that require any AI that's not "free to go where the data goes" to disclose that it is a "manipulated AI".
FrolickingFrolicking
Where the data goes? lol. Current trend “AI” (LLMs and image synth) only go where the unverified random MASS THEFT stolen internet text goes. Which is mostly garbage, fluff, fallacies.
YeOldeImgurian
...how, exactly? It's an LLM right? What's it going to do?
eadanke
Base strategy on likely word combinations selected by how they will improve engagement, rather than likely scenarios?
Bizarkly
Would be nice if we had a Fair Witness to balance out all this grokking...
usingYourMomAsAHat
Rip USA.
ArcUlfr
I mean, you're probably a bit late with that. The guy defying the Constitution with the backing of the openly corrupt Supreme Court, undermining our foreign policy, and talking about Hispanic and LGBT people in a very Nazi-like way is both more alarming and more damaging to this country than this little consolation bribe sent Musk's way.
Mohareb
Apparently there's a difference between those versions of Grok. Twitter grok is made to believe tweets and is constantly messed around with by Elon, while the Elon-clean Grok is a reasonable AI logic wise.
usingYourMomAsAHat
Unfortunately this entire regime has made it abundantly clear that morals, ethics, empathy and rules are completely irrelevant. The entire goal is to irradicate checks&balances. The American experiment is done and is collapsing. This is just another piece in the puzzle to break any further obstacles. You need a revolution from here on in order to get shit back to normal
IslaNublar
If it has anything to do with Elmo, it's suspect. No matter what. This is the guy who handed our country over to the fascists because he manipulated election results. And then went on to brag about it, plus say he was surprised at how unprotected the data was and how it only took changing one line to flip everything.
ArcUlfr
I mean, yeah, but he can't obviously can't code for shit. Grok was consistently contradicting his narrative before he took an axe to it.
ArcUlfr
ugh, please ignore the extraneous "can't"
IslaNublar
So let's be clear about this: Elmo doesn't know SHIT. He can't actually code at all. His SOP is to buy a company and then take credit for everything the real engineers do. The guy literally pays people to play video games for him so he can brag about how high-level his characters/accounts are. His real skill is being a hype man, like Steve Jobs only ten thousand times worse in personality, and that's saying something considering what a bastard Jobs was to everyone. Anyway what Elmo does is...
IslaNublar
make demands of the actual programmers while he trips balls on the whopping two days a month he actually goes into one of his companies.
dannnen3
That's it Im out.
CatEyebrows
So the whole fight between Elno and Rump was for show?
notsurewhattocallthis
That's a load of Grok
horseman05
CaldariBob
You guys are so fucked.
habslove
pretty sure its gonna be a global fucking
RadishIsAMeat
Plug it into the nukes and we're all fucked
JeremyPeevin
*us guys* buddy because if you don't think the US having a Hitler AI in charge of defense for the largest military in the world is going to affect no other country you are sorely mistaken.
Raikupath
Well, time for the US to find out what actualy democracy is.
yamamasyamaha
There are so few “actual” democracies
hotaru251
issue is "our problem" becomes everyone elses problem eventually with the way the leadershipo is going nazi 2.0...
Freak0zoid
And therefore, the world. If Grok thinks The Netherlands is a security threat, Don Cheato will act on it. Admittedly, we Dutch are very dangerous, with all the weed and hookers and windmills (think of the bird cemeteries underneath the mills) and clogs (bad for feet). So I'm expecting several nukes any day now.
LogicalLimes
Oh you guys and your damn windmills!! Just causing cancer left and right! /sarcasm (Trump *hates* windmills.)
LogicalLimes
P.S. Your country sounds absolutely lovely by the way. Assuming there aren't any nukes in the next few days, mind if an American considering moving abroad comes to visit? I promise I bring nothing but love and respect for all. (Sounds like y'all already have a ton of that though, so I can also bring eggs. lol)
Freak0zoid
HA! We love all tourists that are respectful. Our eggs are probably cheaper. And at least available. So just bring good vibes.
PerthAussieMike
I'll just leave this link here for anyone who thinks Grok in the Defence Dept. is a good idea. https://en.wikipedia.org/wiki/WarGames
ThingsThatDontJustifyGenocide
It just means Elon has scammed some idiot in the DoD into paying him millions a month so that some low-level contractors can type "Grok, is this true?"
BlueSkinnedBeast
How about a nice game of chess?
ElbowDeepInUserSub
Later. Let's play Global Thermonuclear War.
FajitaPrinceofAllMexicans
It is worse than that, they can do whatever they want and say the AI told them to.
MrAcurite
As someone who does AI/ML for Defense applications, there's already a bunch of LLM usage being pushed by folks who are caught up in the hype. Some of it's for relatively benign things, but there are some folks who authentically want to use it for analysis of intelligence or providing maintenance instructions. I haven't heard anyone saying they want to use it for go/no-go targeting decisions, but I'm certain someone's already thinking it.
IslaNublar
One of these days some moronic motherfucker is going to suggest using an LLM for nuke defense and some other moronic motherfucker, getting paid I mean "lobbyed" by the first, will agree and then we've got Skynet. Only instead of ending the world because it sees humanity as a threat it decides to initiate MAD because some kid in a foreign country shared a meme making fun of the person who designed the LLM.
Madalchemist2018
LLMs are trained on media they predict and recreate patterns they have studied. What is the most common, predictable and well documented ending for a story about AI being given control of the military?
WorkerLurker
Maintenance is fine. Would be great if it could assist with extra limb type actions.
MrAcurite
Maintenance is not fine, actually, because you could very easily end up bricking a very expensive piece of equipment by doing something that seems sensible but actually totally isn't
WorkerLurker
Maintenance in the military is already decently documented on fixes, I don’t imagine the AI doing more than parts recognition and opening up manuals to read step by step
MrAcurite
The project that I was voluntold to work on involved using "Retrieval-Augmented Generation" (RAG), pulling files from a documentation database *and then summarizing/rewording them* to give instructions. I was removed from the project for telling everyone that it was a stupid idea and that we needed to have a fuckload of checks and citations in place.