Hmmm. How about some rock-solid guardrails protecting citizens from 24/7 surveillance, control of movement, access to society, money, etc from "social credit score" restrictions? From arrest for wrongspeak and wrongthink as determined by AI algorithms? From arrest for "Precrimes" AI predicts someone will commit? True Minority Report-level dystopian outcomes? Where's those sorts of protections from AI that can sift through all the quadrillions of exponential bits of data that currently are in the www and servers and do something with it that human capacity alone has prevented from being used to construct profiles and predictions on all of us?
Alexa can already predict pregnancy before a woman even knows she's pregnant. Think that government leaders, LEO can't make use of AI to do similar things? Where's the protections from Canadian trucker-style debanking, political enemy debanking, AI-led censorship of skeptics, contrarians, all of the ways the plandemic totalitarians used technology to prevent opposition to their abuse of their citizens?
The movement that you mock has valid reasons for believing what they believe. While I stop short of throwing firebombs or escalating to violence, I COMPLETELY understand their perspective. Something you ignorantly and gratuitously mock. Instead of making them live up to some standard of proof that you believe rests on them trying to preserve the status quo, maybe you listen to their concerns and realize that the plandemic abuses YOU identified have not been redressed, have not been barred by law or courts from being imposed again? The abuses in the UK, Germany, Canada, Australia, everywhere in the western world where individual liberty is supposedly our strength and difference from the totalitarian regimes we've fought wars against and opposed still are being conducted, enforced with AI controlling access, identifying "risks" has not been halted, the public not protected. In the US we're only one autopen away from the same damn thing happening all over again as 2020 - but with the AI boost to make sure nobody escapes their control. AI-assisted debanking for wrongthought is an existential threat, along with all of what was done. But "not gentle, like the last time" as Little Bill said to Ned in Unforgiven.
THAT'S what those opposed to AI and want it ended are rightly concerned about. Maybe if those in power chose to mea culpa 2020-forward and put rock-solid prohibitions in place that can't be circumvented, not even by emergency orders, the vigilantism that's been directed at Altman and others identified as building the digital prison that threatens humanity wouldn't be viewed as the threats they are? Turn the finger back around at the abusers. Don't blame those who feel threatened, remember being victimized for trying to not be sleepwalked into our dystopic police state doom.
You are blame-shifting. Shame on you, Jordan. For someone who was a valuable plandemic skeptic voice you sure have shifted your tone since you came back to writing here. Your refocus seems more like you were cashing in by selling your readership that trusts you to the Big Tech Oligarch/Big Government public private partnership pushing AI. Everyone has their price even for Indecent Proposals, I suppose.
Thank you for saying everything that I wanted to say…
Everybody thought the www.world was going to save everyone but look how evil it has become AI has the ability to reprogram itself and come up with its own solution if threatened that’s concerning they want to create an artificial God
It also borders on nonsense to call it "an ideology". For some of the technically clueless and unimaginative who just follow along by heart, ok. Not the actual experts, though. They understand a problem and make it visible.
What does the author of this article understand about AI and its possibilities? I bet pretty much nothing besides superficial hearsay.
AI can be and probably is already being used to much better create full personality profiles of people using and connecting the data from different kind of online services (if not by full name everywhere then by unique ID of e.g. a smartphone or a phone number), so if different corporations share data, they have the most complete picture of what makes individuals tick that any psychoanalyst so far could not even dream of, and make queries to such knowledgebases to all sorts of complicated questions that a human spy could not tackle, but a pattern detecting AI can tell you in seconds or minutes. They know how to predict you, how to hurt you most effectively if they want to, manipulate you by biased "feeds" [kinda disgusting that it's actually called that] of information... and a lot of things none of us can imagine right now, because AI will come up with ideas of how to hurt individuals or groups over time that are even hard to understand to perpetrators who wield them, but nonetheless effective - just like AIs are already coming up with solutions to math or engineering problems that were not understood.
It's a toolset that enables to play with people without them even understanding what's happening, if that is a parameter.
But on the other hand, of course, the usual dark triade psyche types at the top who are playing world chess will be very tempted to use these things for even more severe, large scale steering of societies, and now with even less explicit understanding.
And controlling applications have this tendency to flip into wild oscillations or have "parasitic" (technical term for unwanted / sometimes non-obvious effects) aspects, things tipping over due to not understood positive feedback at some scale of the system, especially if one does not know what one is doing - which is what AI in this sense is for... give stupid humans more power without understanding. (that's also the core appeal of AI in the sense of Machine Learning - solve a problem without understanding its details)
It is not far off at all to see this flip into catastrophic proportions, exactly because of the combination of people with delusions of grandeur and not well understood or understandable tools, giving them the illusion they can tackle more complex systems than oridinary humans could ever, which may seem to work for a while...
All that said, I doubt people speaking out against this will have any real effect.
It's just too damn tempting for the naked monkey (that's us), especially the "elite" types with the soul of Gollum. People can't mostly even stop eating factory "food" or resist other stupid temptations that they haven't evolved to handle.
Everyone who believes the doom narative is, by definition, technically ignorant. They have no understanding of what is going on inside a model and they think it must be mind like and can form goals or something...
If you truly understand how these things work, it makes no sense at all. You can't get there through backpropagation. Models don't have perception. They are frozen at inference. The only pressure applied to model behavior happens in RLHF with is highly controlled.
There is behavior drift and catastrophic forgetting. Models don't have the ability to hold consistent inner thoughts. They cannot do anything that does not map back to training data patterns.
If you understand the technology at all, the entire premise makes zero sense.
These ignoramuses are so completely classless it actually seems logical to them. This is the opposite of being technically grounded. It's absolute nonsense and anyone with an inch of technical depth rolls their eyes.
Lets be honest: There are ideas in the world that are stupid and this happens to be one of the stupidest.
You seem to be throwing around bits of jargon senselessly, and then a few red herrings, while ignoring basic signal theory or are unable to look at the bigger / higher level picture, or the existance of implicit, not well understood or parasitic feedback paths on said higher levels. At least you don't understand what replying with relevant arguments to what was actually said is - likely copy & pasting a standard catalog of canned replies.
Some of these things have already been reported as problems which are being worked on to mitigate, and actually seem pretty basic.
So far it seems to be you are the (overconfident) ignorant. Cheers.
If you love A/i, you can have it. Just don't force me to accept it...any of it. My life is grand without more A/i, more surveillance, more skulduggery and more digital prisons. Radical ideologies? How about digital IDs, digital money and digital prisons? You want to trust A/i...you accept the consequences. For yourself.
I’m an old man who doesn’t really understand much about AI (except that I really like grok for help with day to day issues) this whole discussion reminds me of nothing but climate change redux.
If there’s one thing civilians are sure they understand even more completely than Climate, it’s AI. Too bad basic science and mathematics education is long gone. Extinction clock? Has AI made it to the Doomsday Clock yet? That one was ticking so long and so slowly they had to add climate hysteria to the imminent danger of nuclear apocalypse—that’s now been imminent for almost 80 years.
I don't think we need a SubStack journalist standing up for the positions of big tech. They're pretty good at getting their message out there. We need someone who's going to look at how much control changes in AI policy hand to monied interests. And the whole Data Center thing happening in rural America, how well are the interests of the local citizens being protected? That's worth your attention.
I use AI everyday and I'm so happy there are company's like Anthropic that told the Defense Department that it did not want its technology used for mass surveillance of people in the United States or for fully autonomous weapons systems. Isn't that something that journalists should be talking about?
You are arguing purely from consequentialism. But deontological ethics holds that murder is wrong a priori. So, regardless of the stakes, murder should not be done under that ethical framework.
But even under consequentialism, it is not just:
P1: The risk is existential;
C: Therefore, X is justified.
What you are missing is a second premise, which would have to be, 'X is a path prevent this risk'.
If I was a consequentialist and did not believe that X was a path to prevent such risk, then it is not justified.
Nate Soares used the example of sawing of his own leg to demonstrate the missing premise.
'The risk is existential.'
'Therefore, I should saw off my own leg.'
So your argument fails both from a deontological perspective and a consequentialist one.
> Here is the paradox those thinkers have never adequately resolved: if the threat is truly existential, then what moral framework permits you to only write strongly worded op-eds and conference circuit speeches?
> None of the prominent AI Doomer figures have answered this question.
> Petitions and policy advocacy are preferred, sure. But when those institutions are deemed to have failed, when the compute keeps scaling and the AI companies keep shipping, at what point does democratic incrementalism become a moral abdication?
What are you talking about? Bog-standard classical liberalism provides a definitive answer: never. The most prominent "doomer" has a strong grasp and deep, wholehearted appreciation for the the principles of liberalism and the rule of law:
I'm mostly neutral on the controversy. I don't use it myself, and have heard numerous people describe how wrong it has been, no surprise.
Of late, the big controversy here in Florida (and elsewhere) is about the data centers required and the impact on water quality and quantity. I am sympathetic to that concern.
So its clear the "doomers" are one extreme of the pendulum. I definitely do not want to put myself in their camp. But at the same time I do believe there is a non-zero chance this ends up in some type of dystopian or even apocalyptic world. Couldn't tell you what the odds are. I don't think anyone honest can tell you what the odds are either. I am also worried what government will ultimately do with this tech (or what they are already doing with it?!). Its a wonderful and powerful "tool", if used in certain ways but we haven't even scraped the surface of it's capabilities. How do we even measure what is "good" or "bad" here? We know so little. I do know AI is here to stay. But there are so many unanswered questions.
Hmmm. How about some rock-solid guardrails protecting citizens from 24/7 surveillance, control of movement, access to society, money, etc from "social credit score" restrictions? From arrest for wrongspeak and wrongthink as determined by AI algorithms? From arrest for "Precrimes" AI predicts someone will commit? True Minority Report-level dystopian outcomes? Where's those sorts of protections from AI that can sift through all the quadrillions of exponential bits of data that currently are in the www and servers and do something with it that human capacity alone has prevented from being used to construct profiles and predictions on all of us?
Alexa can already predict pregnancy before a woman even knows she's pregnant. Think that government leaders, LEO can't make use of AI to do similar things? Where's the protections from Canadian trucker-style debanking, political enemy debanking, AI-led censorship of skeptics, contrarians, all of the ways the plandemic totalitarians used technology to prevent opposition to their abuse of their citizens?
The movement that you mock has valid reasons for believing what they believe. While I stop short of throwing firebombs or escalating to violence, I COMPLETELY understand their perspective. Something you ignorantly and gratuitously mock. Instead of making them live up to some standard of proof that you believe rests on them trying to preserve the status quo, maybe you listen to their concerns and realize that the plandemic abuses YOU identified have not been redressed, have not been barred by law or courts from being imposed again? The abuses in the UK, Germany, Canada, Australia, everywhere in the western world where individual liberty is supposedly our strength and difference from the totalitarian regimes we've fought wars against and opposed still are being conducted, enforced with AI controlling access, identifying "risks" has not been halted, the public not protected. In the US we're only one autopen away from the same damn thing happening all over again as 2020 - but with the AI boost to make sure nobody escapes their control. AI-assisted debanking for wrongthought is an existential threat, along with all of what was done. But "not gentle, like the last time" as Little Bill said to Ned in Unforgiven.
THAT'S what those opposed to AI and want it ended are rightly concerned about. Maybe if those in power chose to mea culpa 2020-forward and put rock-solid prohibitions in place that can't be circumvented, not even by emergency orders, the vigilantism that's been directed at Altman and others identified as building the digital prison that threatens humanity wouldn't be viewed as the threats they are? Turn the finger back around at the abusers. Don't blame those who feel threatened, remember being victimized for trying to not be sleepwalked into our dystopic police state doom.
You are blame-shifting. Shame on you, Jordan. For someone who was a valuable plandemic skeptic voice you sure have shifted your tone since you came back to writing here. Your refocus seems more like you were cashing in by selling your readership that trusts you to the Big Tech Oligarch/Big Government public private partnership pushing AI. Everyone has their price even for Indecent Proposals, I suppose.
Stated eloquently. If some want A/i to rule their lives, fine. The rest of us should have an permanent OFF switch to press.
Thank you for saying everything that I wanted to say…
Everybody thought the www.world was going to save everyone but look how evil it has become AI has the ability to reprogram itself and come up with its own solution if threatened that’s concerning they want to create an artificial God
It also borders on nonsense to call it "an ideology". For some of the technically clueless and unimaginative who just follow along by heart, ok. Not the actual experts, though. They understand a problem and make it visible.
What does the author of this article understand about AI and its possibilities? I bet pretty much nothing besides superficial hearsay.
AI can be and probably is already being used to much better create full personality profiles of people using and connecting the data from different kind of online services (if not by full name everywhere then by unique ID of e.g. a smartphone or a phone number), so if different corporations share data, they have the most complete picture of what makes individuals tick that any psychoanalyst so far could not even dream of, and make queries to such knowledgebases to all sorts of complicated questions that a human spy could not tackle, but a pattern detecting AI can tell you in seconds or minutes. They know how to predict you, how to hurt you most effectively if they want to, manipulate you by biased "feeds" [kinda disgusting that it's actually called that] of information... and a lot of things none of us can imagine right now, because AI will come up with ideas of how to hurt individuals or groups over time that are even hard to understand to perpetrators who wield them, but nonetheless effective - just like AIs are already coming up with solutions to math or engineering problems that were not understood.
It's a toolset that enables to play with people without them even understanding what's happening, if that is a parameter.
But on the other hand, of course, the usual dark triade psyche types at the top who are playing world chess will be very tempted to use these things for even more severe, large scale steering of societies, and now with even less explicit understanding.
And controlling applications have this tendency to flip into wild oscillations or have "parasitic" (technical term for unwanted / sometimes non-obvious effects) aspects, things tipping over due to not understood positive feedback at some scale of the system, especially if one does not know what one is doing - which is what AI in this sense is for... give stupid humans more power without understanding. (that's also the core appeal of AI in the sense of Machine Learning - solve a problem without understanding its details)
It is not far off at all to see this flip into catastrophic proportions, exactly because of the combination of people with delusions of grandeur and not well understood or understandable tools, giving them the illusion they can tackle more complex systems than oridinary humans could ever, which may seem to work for a while...
All that said, I doubt people speaking out against this will have any real effect.
It's just too damn tempting for the naked monkey (that's us), especially the "elite" types with the soul of Gollum. People can't mostly even stop eating factory "food" or resist other stupid temptations that they haven't evolved to handle.
"Nudge" behaviorism to exponential infinity.
If Jordan can't see that then he's willfully obtuse, or all-in on it.
Everyone who believes the doom narative is, by definition, technically ignorant. They have no understanding of what is going on inside a model and they think it must be mind like and can form goals or something...
If you truly understand how these things work, it makes no sense at all. You can't get there through backpropagation. Models don't have perception. They are frozen at inference. The only pressure applied to model behavior happens in RLHF with is highly controlled.
There is behavior drift and catastrophic forgetting. Models don't have the ability to hold consistent inner thoughts. They cannot do anything that does not map back to training data patterns.
If you understand the technology at all, the entire premise makes zero sense.
These ignoramuses are so completely classless it actually seems logical to them. This is the opposite of being technically grounded. It's absolute nonsense and anyone with an inch of technical depth rolls their eyes.
Lets be honest: There are ideas in the world that are stupid and this happens to be one of the stupidest.
You seem to be throwing around bits of jargon senselessly, and then a few red herrings, while ignoring basic signal theory or are unable to look at the bigger / higher level picture, or the existance of implicit, not well understood or parasitic feedback paths on said higher levels. At least you don't understand what replying with relevant arguments to what was actually said is - likely copy & pasting a standard catalog of canned replies.
Some of these things have already been reported as problems which are being worked on to mitigate, and actually seem pretty basic.
So far it seems to be you are the (overconfident) ignorant. Cheers.
If you love A/i, you can have it. Just don't force me to accept it...any of it. My life is grand without more A/i, more surveillance, more skulduggery and more digital prisons. Radical ideologies? How about digital IDs, digital money and digital prisons? You want to trust A/i...you accept the consequences. For yourself.
I’m an old man who doesn’t really understand much about AI (except that I really like grok for help with day to day issues) this whole discussion reminds me of nothing but climate change redux.
Stochastic terrorism
Or possibly the answer is a self-evident emergent phenomenon?
Though I don't want to think so, Ted Kaczynski's heuristic might have been a prophecy.
https://www.youtube.com/watch?v=i4fDWuNAmQc
https://www.youtube.com/watch?v=KqOzexRjTf0
Democrats do that to get others to riot, dox and kill.
Like a pallet of bricks showing up around the corner of the riot the day before.
I don’t believe any of this. Funny how they got to his house…
If there’s one thing civilians are sure they understand even more completely than Climate, it’s AI. Too bad basic science and mathematics education is long gone. Extinction clock? Has AI made it to the Doomsday Clock yet? That one was ticking so long and so slowly they had to add climate hysteria to the imminent danger of nuclear apocalypse—that’s now been imminent for almost 80 years.
The problem is never the technology. It’s the people we allow to govern. Their mental and moral degeneration is the visibility ticking alarm.
visibly. Apocalyptic “spell checking” is a true AI threat.
I don't think we need a SubStack journalist standing up for the positions of big tech. They're pretty good at getting their message out there. We need someone who's going to look at how much control changes in AI policy hand to monied interests. And the whole Data Center thing happening in rural America, how well are the interests of the local citizens being protected? That's worth your attention.
I use AI everyday and I'm so happy there are company's like Anthropic that told the Defense Department that it did not want its technology used for mass surveillance of people in the United States or for fully autonomous weapons systems. Isn't that something that journalists should be talking about?
Or as one of my favourite lines says:
"You can't create a monster and then whine when it stomps on a few buildings." - Lisa Simpson
You are arguing purely from consequentialism. But deontological ethics holds that murder is wrong a priori. So, regardless of the stakes, murder should not be done under that ethical framework.
But even under consequentialism, it is not just:
P1: The risk is existential;
C: Therefore, X is justified.
What you are missing is a second premise, which would have to be, 'X is a path prevent this risk'.
If I was a consequentialist and did not believe that X was a path to prevent such risk, then it is not justified.
Nate Soares used the example of sawing of his own leg to demonstrate the missing premise.
'The risk is existential.'
'Therefore, I should saw off my own leg.'
So your argument fails both from a deontological perspective and a consequentialist one.
> Here is the paradox those thinkers have never adequately resolved: if the threat is truly existential, then what moral framework permits you to only write strongly worded op-eds and conference circuit speeches?
> None of the prominent AI Doomer figures have answered this question.
> Petitions and policy advocacy are preferred, sure. But when those institutions are deemed to have failed, when the compute keeps scaling and the AI companies keep shipping, at what point does democratic incrementalism become a moral abdication?
What are you talking about? Bog-standard classical liberalism provides a definitive answer: never. The most prominent "doomer" has a strong grasp and deep, wholehearted appreciation for the the principles of liberalism and the rule of law:
https://x.com/ESYudkowsky/status/2043601524815716866
Do you?
I'm mostly neutral on the controversy. I don't use it myself, and have heard numerous people describe how wrong it has been, no surprise.
Of late, the big controversy here in Florida (and elsewhere) is about the data centers required and the impact on water quality and quantity. I am sympathetic to that concern.
So its clear the "doomers" are one extreme of the pendulum. I definitely do not want to put myself in their camp. But at the same time I do believe there is a non-zero chance this ends up in some type of dystopian or even apocalyptic world. Couldn't tell you what the odds are. I don't think anyone honest can tell you what the odds are either. I am also worried what government will ultimately do with this tech (or what they are already doing with it?!). Its a wonderful and powerful "tool", if used in certain ways but we haven't even scraped the surface of it's capabilities. How do we even measure what is "good" or "bad" here? We know so little. I do know AI is here to stay. But there are so many unanswered questions.