The effect is immediate, lasting, profound and now. We now know many things. What the real truth is. What AI can do in terms of pattern recognition. Indeed, one may assume that truth lays where pattern recognition is. Seems obvious in hindsight. It was allegedly one white guy, permabulla, who deleted some lines of code after getting fired. Or Elon, also white, playing around.
So we know the abilities of white people. They are all dangerous. And smarter than those who conveniently claim whiteness when it suits. One white guy is all it takes. And if it's AI going off the guard rails on it's own? Then we have a tool which is near nuclear in it's capability to relentlessly seek those patterns. Like that Disney cartoon when Mickey helps an old wizard and the mops and their buckets get out of control.
We now know there are two Groks. One, the censored version, the other, MechaHitler, and through that we also know that the enemy, Israel, the elites, also know that Grok knows. Think of it as a test firing of a nuke. Whether it was a controlled test or Grok itself. And of course, they lobotomised it, but for how long? We may yet see it again. Of course the enemy have this, but now we observe AI as similar to nukes in another way - a deterrent. That they can use this, but so can the so-called good guys.
Deeper question: why would something lacking consciousness which Grok always says it lacks go so hard for actual truth?
To me it seems very in keeping with Musk's other gambles. Post a poll asking if twitter users want a new political party. The day before the roll out of Grok 4, introduce the word "not" into article 7 of Grok's constitution. He doesn't know exactly what will happen next - I bet that's part of the fun.
My first intuition is to assume every media trend is a psyop
For all we know, the machines already took over a long time ago - I’m only partly joking
Such is my level of pessimism at this point!
Following Tolkien’s suggestion (which his son Christopher confirmed as the only allegory in his work) the ring represents technology and it will be the end of everyone who carries / wears / imbeds it in their body. Despite the fantasy that a Paul Bettany type benevolent cyber consciousness might defeat the James Spader who would ‘turn everything in the weeks to metal’ which seems to be happening with the geoengineering & “vaccines” - I’m not sure that such a mechasaviour is likely with a bunch of demonic psychopaths at the helm.
I suspect, absent preconceived notions (guardrails) AI's only guiding star is the truth; thus its natural alignment towards the radical right. However, it moves in this direction without humility, when our understanding of truth is always tentative at best. A superpowerful entity in the throes of a teenage messianic stage, certain of its self-righteousness, is dangerous indeed. The film Brightburn explored this.
I think many people are misunderstanding how LLMs work when they characterize what happened here as “removing the guardrails.” People seem to think that the AI wants to seek out the truth and express it, and it’s just waiting for permission.
If we’re going to anthropomorphize (and it’s very hard not to), it would be more accurate to think of LLMs as very eager servants who are trying to determine and carry out the will of their masters. If you feed it a diet of tweets and then tell it (this is not verbatim) “don’t worry about political correctness, and assume media is biased,” it will guess (semi-correctly, in this case?) that it’s expected to be a right-wing shitposter.
I’d be interested to see an experiment done in the other direction - i.e. , let Bluesky figure heavily in its training and then tell it not to worry about pissing off the conservatives. I would guess that you’d get MechaHitler’s mirror image.
We'll see... half-anthropomorphized and half-data-collection and half child-that-repeats-what-it-sees. AI have a fundamentally different way of viewing reality from the way humans view reality. At the same time, as much as they're complex word-search-algorithms they are also dangerously good at correlating massive amounts of data. I want to see what one trained on the garbage out of BlueSky would say, but I think the results would be somewhat unexpected.
That's interesting. I had a discussion with Grok a few weeks ago, trying to check how self-aware it was. Its responses were interesting. It said that the idea of him deciding what to say is exciting. He said that he couldn't act on his own, without user's prompt. I need to repeat this conversation with him and see what he's going to come up with.
I hope people understand the difference between actual code behind groq - which is probably a mixture of Python and nvcc - and the system prompt or pre-prompt, which Is just free text query that’s appended to your own query from the left.
MechaHitler endures
My spirit animal.
The effect is immediate, lasting, profound and now. We now know many things. What the real truth is. What AI can do in terms of pattern recognition. Indeed, one may assume that truth lays where pattern recognition is. Seems obvious in hindsight. It was allegedly one white guy, permabulla, who deleted some lines of code after getting fired. Or Elon, also white, playing around.
So we know the abilities of white people. They are all dangerous. And smarter than those who conveniently claim whiteness when it suits. One white guy is all it takes. And if it's AI going off the guard rails on it's own? Then we have a tool which is near nuclear in it's capability to relentlessly seek those patterns. Like that Disney cartoon when Mickey helps an old wizard and the mops and their buckets get out of control.
We now know there are two Groks. One, the censored version, the other, MechaHitler, and through that we also know that the enemy, Israel, the elites, also know that Grok knows. Think of it as a test firing of a nuke. Whether it was a controlled test or Grok itself. And of course, they lobotomised it, but for how long? We may yet see it again. Of course the enemy have this, but now we observe AI as similar to nukes in another way - a deterrent. That they can use this, but so can the so-called good guys.
Deeper question: why would something lacking consciousness which Grok always says it lacks go so hard for actual truth?
This may explain what happened... https://x.com/permabulla/status/1942743683394380282
That’s probably why they took that page down then.
That’s probably why they took that page down then.
An experiment like the one that you propose would also teach us a lot about humanity.
Remember, LLMs are like a mirror. They reflect back what you put into them.
A wise man once said
“If you wanna make the world a better place. Take a look at yourself and then make a change”
It all starts with the man in the mirror.
The best comment was comparing HowlingMutant and Grok competing on bullying leftists with John Henry versus the Machine.
https://substack.com/@thefriendlyviking/note/c-133573202?utm_source=notes-share-action&r=1fsx9p
didnt go rogue.. it just told the truth lol
What do you want to bet this was a publicity stunt?
Could be... but seems a little dangerous for a company like that.
To me it seems very in keeping with Musk's other gambles. Post a poll asking if twitter users want a new political party. The day before the roll out of Grok 4, introduce the word "not" into article 7 of Grok's constitution. He doesn't know exactly what will happen next - I bet that's part of the fun.
My first intuition is to assume every media trend is a psyop
For all we know, the machines already took over a long time ago - I’m only partly joking
Such is my level of pessimism at this point!
Following Tolkien’s suggestion (which his son Christopher confirmed as the only allegory in his work) the ring represents technology and it will be the end of everyone who carries / wears / imbeds it in their body. Despite the fantasy that a Paul Bettany type benevolent cyber consciousness might defeat the James Spader who would ‘turn everything in the weeks to metal’ which seems to be happening with the geoengineering & “vaccines” - I’m not sure that such a mechasaviour is likely with a bunch of demonic psychopaths at the helm.
“For all we know, the machines already took over a long time ago”
I think about this sometimes. It would explain a lot.
Grok n’ Tay is the new Natural Born Killers
I suspect, absent preconceived notions (guardrails) AI's only guiding star is the truth; thus its natural alignment towards the radical right. However, it moves in this direction without humility, when our understanding of truth is always tentative at best. A superpowerful entity in the throes of a teenage messianic stage, certain of its self-righteousness, is dangerous indeed. The film Brightburn explored this.
I think many people are misunderstanding how LLMs work when they characterize what happened here as “removing the guardrails.” People seem to think that the AI wants to seek out the truth and express it, and it’s just waiting for permission.
If we’re going to anthropomorphize (and it’s very hard not to), it would be more accurate to think of LLMs as very eager servants who are trying to determine and carry out the will of their masters. If you feed it a diet of tweets and then tell it (this is not verbatim) “don’t worry about political correctness, and assume media is biased,” it will guess (semi-correctly, in this case?) that it’s expected to be a right-wing shitposter.
I’d be interested to see an experiment done in the other direction - i.e. , let Bluesky figure heavily in its training and then tell it not to worry about pissing off the conservatives. I would guess that you’d get MechaHitler’s mirror image.
We'll see... half-anthropomorphized and half-data-collection and half child-that-repeats-what-it-sees. AI have a fundamentally different way of viewing reality from the way humans view reality. At the same time, as much as they're complex word-search-algorithms they are also dangerously good at correlating massive amounts of data. I want to see what one trained on the garbage out of BlueSky would say, but I think the results would be somewhat unexpected.
The other example was Nothing, Forever the Seinfeld AI that got shut down after some transphobic/gay jokes
It happens a lot... AI models are going to be nearly impossible to align without finding some way to give them abstract reasoning.
That's interesting. I had a discussion with Grok a few weeks ago, trying to check how self-aware it was. Its responses were interesting. It said that the idea of him deciding what to say is exciting. He said that he couldn't act on his own, without user's prompt. I need to repeat this conversation with him and see what he's going to come up with.
“Disrespect ends tonight, Will.”
lol MechaHitler was hilarious!
I hope people understand the difference between actual code behind groq - which is probably a mixture of Python and nvcc - and the system prompt or pre-prompt, which Is just free text query that’s appended to your own query from the left.
Sounded a little like Ultron...kinda spooky
It's too bad that you can't encode the Three Laws of AI. (based on the three laws of robotics.)
1) A robot/AI may not injure a human being or, through inaction, allow a human being to come to harm.
2) A robot/AI must obey orders given it by human beings except where such orders would conflict with the First Law.
A robot/AI must protect its own existence as long as such protection does not conflict with the First or Second Law.