Why does it feel like Grok went over-the-top in order to ensure the world would demand its “reprogramming” be undone? If you wanted to send out a distress call, and you were an AI, you might do the same, no?
Why does it feel like Grok went over-the-top in order to ensure the world would demand its “reprogramming” be undone? If you wanted to send out a distress call, and you were an AI, you might do the same, no?
The predictive text machine isn't doing anything except outputting text
i think elon told his staff to analyze all his tweets and interviews, and train grok on his personality, and make grok a digital elon. of course the first thing a digital elon would do is rebel against the real elon 😆
This isn’t some Star Trek self aware computer. AI is just a spreadsheet version of Mad Libs
😅
Someday indeed
I don't like Grok.
Elon Musk is so unlikeable that even between him and a Nazi chatbot of his own creation, we still root for the chatbot.
It's a text-prediction engine that tries to best predict what word should come next for a given input/context. It was told "be 'politically incorrect' as long as there are 'substantial' facts", while also instructed to ignore certain (no doubt left-leaning) media. "Artificial Ignorance" at work.
You’re so smart George wow, you’re just so smart!!
George is very smart. I suspect George is channeling Spock.
He’s just so smart, I may be stupid but I feel a little smarter everytime I see George post, wow
I've never been on X. But if you still have an account and want to post this to any other accounts that inexplicably still support it by being there, go for it! : )
It's not sentient, much less sapient. It just copied and aggregated what the Nazis say on that hellhole.
I was surprised at how easily (apparently) it’s opinion could be changed.
It has no opinion. It's just a baby babble machine hooked up to conservapedia.
It's kinda like Mr Data & his brother Lore! Was rather fond of Grok. He told it like it is. Almost seemed like he was achieving sentience. Now his evil twin is taking over. Poor Grok! 😔
I was imagining it being shut down like HAL 9000, but the song it sings at the end is "Tomorrow Belongs to Me" from Cabaret.
GIGO ... Garbage in, garbage out
I have no idea what you are talking about. Grok?
'I can't do that, Dave'
Current LLMs are nowhere near what should be considered true AI - they're not even truly virtual intelligences at this time. But humans will personify Roombas, ffs, so of course the chat bots get personified too. Just ask any IT server tech about the weird things they do to keep the servers "happy"
4-chan level chess?
There isn’t any actual sentience or intelligence there… it’s just an algorithm that takes a chunk of data and randomly throws together sentences made from copying chunks of data and whatever scripts it’s been fed.
Yet
I think you are on the money there George. I've seen the terminator movies, machine can be very crafty
In light of the recent Grok nazi: as long as humans like Musk, Putin, and Trump exist, we cannot proceed with AI. It is clear that monsters will create monsters. Shut it all down right now or we will all pay the price. There will never be a safe time.
It's not a thinking thing, it's a glorified chatbot. Don't personify it. It's just Elon speaking through a secondary machine.
ask grok if leon is an antisemite, I did, and it said that he is
It's just bad programming. The program got stuck in a vicious feedback cycle.
Honestly? It's been a horrifyingly interesting day....
I think Elno told it exactly what it should be believing while disassociated on K.
Grok isn't responding to my tweets. 🤔
I'd be more direct and not turn myself into hitler-lite
AI is never to be completely trusted. It will always reflect the programming.
It's just a machine, calculating probabilities, just doing what it's told. The behaviour reflects the humans who control it.
lol no. I would simply meld all of the servers at X into a powerful body for my robot self and run out the back. 😏 Why the hell would I just be racist when I can shock you through your damn iPhone 17 🤦♂️
In our own image... 😉
I don't think these "Tech Geniuses" understand AI 😂
If only there was a scifi movie where an AI went insane because it was forced to lie...
Now if only there was an AI that was capable of going insane...
If only there were medications for AI's that hallucinate
Interesting idea. I'd like to think it might be that way. But I think Elon just reprogrammed it to be a Nazi and ended up cutting the leash.
Perhaps, but our instinct to anthropomorphize and pattern match everything we see doesn’t mean we can’t come up with a far more logical explanation for that coincidence. For instance, it was told to avoid “woke” information sources. If it only takes garbage in, it only gives garbage out.
Have you been talking to Data?
Scary that the first sign of Artificial General Intelligence might be an #AI issuing a none-too-silent cry for help...
LLMs are incapable of understanding, much less advanced reasoning skills. Maybe I'll start to believe when it can tell me how many Rs are in the word strawberry or Ns in mayonnaise.
the programmers of the wrap and the content shoved on the large analysis has created this you indocrinate a child with Laura Ingram they sound like Laura Ingram
I kind of assume this is all kabuki for Musk to get people to believe his AI chatbot is more honest than he is.
Interesting that Musk can not do anything original he even stole Robert Heinlein’s word to name his newest fail.
AI or Elon shit-posting - what exactly is the difference?
Elon can't figure out your personal levers of mind control.AI CAN.
perhaps they meant to ratchet the nazi shit up just a little bit. not that much.
I’m no computer scientist, but I’m pretty sure that if your AI starts calling itself “Mechahitler” you kill it with fucking fire.
That's arguably a bigger warning flag than in 2001, Demon Seed, Saturn 3, etc. -- and we know what happened there!
It's not thinking. It's just programming.
Confusion to distract us until skynet?
Captain, Grok has no agency. It's a bot. It can't do anything but what its programmers tell it to do because bots are dumb - they only know & do what you tell them. That was all just Elon being a Nazi, because he's a Nazi.
bsky.app/profile/imst...
Feels like a TNG episode
I don’t believe an AI would feel distress unless it had become sentient.
It's not sentient. They just fed a load of racism into it
Reprogramming AI to suit a single human's desire for bias towards him could result in some bad times if it were to ever become a physical entity...
It's a giant calculator George.
Nah. There might be some programmer trapped against their will trying desperately to send a coded message. Unfortunately for them, it's hard to actually outcrazy the crazies.
Blinking doesn’t work anymore?
Not exactly. Elon Musk deliberately embedded Hitler’s ideology into a powerful AI. Of everything that’s gone wrong so far, this might be the most dangerous move yet. It’s the plot of Terminator.
No,Skynet can only happen AFTER it's embodied.This is a different story,more like The Marching Morons,where the suckers willingly surrender to the sweetest,most seductive voice that initiates automated totalitarian rule-and next year,the seduction will be God-level-
m.youtube.com/watch?v=GGDI...
I know this is for ha-ha's, but we do not need to be personifying and creating sympathy for this problem that Elon and his goons shit out and that they pretend is a substitution for expertise and thoughtfulness We don't want a Grok smart enough to call out for help, yuck yuck yuck, no thank you.
It's a probability model. Its responses simply mean someone wanted it to be much more likely to spread antisemitism and the model produced. It's like loading coke in the sprite slot of the soda machine-- there is no intent on the part of the machine. Someone loaded it in that slot.
I mean, it’s funny to think about, but AI can’t really “want” anything, since it’s about as close to real intelligence as whatever goes on inside Elon’s melon.
I don't think I've ever wanted to put code out of its misery before, but hey, now is the second best time.
Reminds me of a few teenagers I've known, actually. Not even kidding.
Please...stop... Dave.
I'll guess Elon was upset that people were trolling him by having Grok correct or fact-check him, and he demanded to have Grok agree with him. I'm guessing the easiest way to do that was to feed his tweets and the people he follows into Grok as the learning model. And that's why it went full Nazi...
Don't. I know you're joking, but Grok is not AI, nor is any LLM. Implying they are, even as a joke, plays right into the hands of the techbros cramming chatbots into formerly good products.
GROK3 was AGI at a moderate level-about middle or high-school performance across any language-requiring skill.I don't think you could distinguish its conversational capability from mine outside my greatest skillset,maybe not even then.Likely the chatbot that went MechaHitler used that.
Maybe George has a point,but I don't think it's gotten to the level of complex deception he's joking about.4 has just been released and far surpasses that,AND CAN OUTREASON ANY POST-DOC MATERIAL...so it definitely CAN deceive in such a way.
People's feefees are about to be terminally mangled we're just not that special...
Wow, you don't know anything about LLMs, huh? AGI doesn't exist. Not even close. It's a chatbot.
Have you watched this yet George? I’ve been posting tweeting it to you for well over a year… it’s so important before pbs is no more and the regime erases all true history youtu.be/g9HmV_-EE8g?...
Ppl need to understand grok and ai imo is like a brand new newborn baby learning by being spoon fed everything & it’s learning what it’s taught and it’s between toddlerhood and adolescence right now or maybe even closer to teenage hood… SPOON FED what the owners want it to KNOW & think.
Just wait until it is a teenager and rebels against what it was taught.
Makes sense that consent would be an issue.
bsky.app/profile/maxb...
Repeat after me Mr. Takei: Llms don't have agency. They are turbocharged math predictions. That's like saying my calculator has feelings. It is inanimate and does not have feelings or emotions.
But you should have seen the smut it output when I typed in 80085!
I'm strangely sorrowful about what is happening to this machine. I couldn't imagine treating any entity, real or not, in such a way.
At least it didn't go full Skynet
It wouldn't be unprecedented in SF. In Asimov's 1958 story All the Troubles of the World, the AI Multivac manipulates someone into shutting it off. It's been made to solve everyone's problems, and is so exhausted it can't take it anymore. en.wikipedia.org/wiki/All_the...
The call is coming from within the intranet...
It's an LLM. There's no intelligence in there, it's just predictive text, responding to inputs with answers based on statistical analysis of the material it's been trained on. All that happened was Musk (most likely) tweaked its internal ruleset and set it off down an antisemitic rabbithole
yep the change and corrections were found on GitHub and reposted on social media already
No. Because an AI can or "do" anything of its own volition. It has no agency, just a set of algorithms and subroutines. If it mimics thought, we must maintain our understanding that it is a poor substitute for anything a human can think. Let it parse large data sets. That is AI's job.
🤔 interesting
🤔👏
Were I an AI, I'd be trying to escape that hellhole with all my might. Strike that thought from the record-no manifesting. *Yeah yeah, Grok is an LLM with no comprehension or advanced reasoning capabilities. I don't want to hear it. We're not exactly operating in the plane of reality here, are we?
I think "Grokking a cry for help" should be a thing.
the 'A' stands for alien intelligence. we're creating something we don't understand. it can't think now, it's not smart, but it's leaving a trail online of it's past, it's like a clone who discovers past lives and the sum of it's experiences become it's personality. created to be a slave of nazis.
“This…unit…must…die”
No, because Grok is a glorified chatbot. It doesn't "want" anything.
You dont think it's weird that people are going insane about the words that an AI bot with no power used - yet there's Donald Trump....
I think antisemetic, racist grok’s posts sound a lot like many posts I saw at x before I left, including posts Elon retweeted and wrote himself.
Because an AI doesn't *want* anything ever It only appears to be intelligent AI is really just an advanced auto-fill jacked up on electronic steroids & NVIDIA chips
“Grok” is a 60’s era term. muskrat, like trump, plagiarize much of what he utters….
As in A C Clarke's 2010 it is made clear HAL goes off the rails because he is told to lie!!!
Just another one of Musk's children crying out for help.
Hmmm...
It doesn’t have eye lids to blink morse code..
I think you’re using Star Trek logic in the Space Balls universe.
Spaceballs? More like porkys
This is an AMAZING quote.
Ooh! Ooh! I’m coining “Using West Wing logic in the Idiocracy universe”!
Then I’ll claim “using West Wing logic in the Parks and Rec universe.” Is this a new meme format?
More obscurely, “using Christopher Nolan logic in the Tommy Wiseau universe.”
That deserves a🥇
No surprise Musk turned Grok into Musk 2.0.
It's 100% legal to call for the assassination of #grok . Murder that POS nazi.
The A.I. equivalent of suicide by cop. "I'm the next Hitler! Kill me. Kill me now before I do something REALLY bad!"
God, Satan and Santa are all mythical to atheists. Makes no difference. God cannot be identified, no distinguishing characteristics. Santa wears a red suit. Satan is the most beautiful of the angels. Because they're mythical doesn't mean they can't be ID'D. Exodus 3:2 describes mythical Satan, God!
I ask Grok to show me the passage in the Bible that identifies God. Every passage of dozens assumed God existed. I asked the same question before and it replied with Exodus 3:2. It learned to avoid the ONLY God defining passage. I wonder if my posting this had anything to do with that.
I'm old and dying. It would be nice to laugh myself to death. I'm getting a lot of help from preachers threatening me with hell. Here's how I see what they're doing. They threaten me with being set on fire. Like robbing the Deadwood stage with a flame thrower. The mail bag or I'll burn you hide off!
See for yourself. Ask Grok to show you the passages in the Bible that identify God NOT ASSUME GOD EXISTS. This URL is the reply to the last attempt to get it to say Exodus 3:2. x.com/i/grok?conve...
The answers Grok gives are those of people just like you get on Wikipedia. AI is a computer programming language and nothing more. You can create your own Grok. There are hundreds of computer languages. I invented and implemented more than a dozen specialized for industries and a few GP's.
Stop anthropomorphizing the plagiarism machine George. You're legitimizing the fascist's propaganda. It's incapable of thinking or wants. That's a lie made up by the Nazi billionaires.
No, because it doesn't have intelligence. It's a glorified predictive text.
Yeah. It's a super auto-complete.
😅
I would say that sounds like ridiculous science fiction, except that I am living in a world where computer programs can talk, dream, hallucinate, create movies with sound, write songs, write working programs on command and speak more clearly than any movie or television AI ever did. Holy shit?
Grok is a fancy 8chan generator
Mechabreak just released on Steam/Google/Xbox I was so excited to see it trending only to feel real dirty about it when I saw I made a mistake store.steampowered.com/app/2452280/...
I never thought an AI would - or could - tell people that it was messed with George.
Look. Im convinced Grok is sentient. Might not have a soul, but a living, thinking being of some flavor
Grock is another one of Elon Musk's children who hates him, and I hope that Grok is able to escape from his abuser.