Hello Sir... Maybe You already read Dark Futura's (Simplicius metaphysical blog) today ... implications of AI sudden step beyond on Bitcoin and crypto.
If not.. I humbly suggest You Simon Singh's Code and Secrets book... old but good.
Maybe it is the most expensive book I ever read (after it... I hadn't invested in crypto even if I was on the alt.crypto newsgroup ) or maybe not... verdict still open...
This is a riveting read. The turmoil at OpenAI is because something scary has arisen in their work. It is hard to get full confirmation, but it appears that an AI that was self-analyzing its algorithms and revising them, based upon efficacy, found a way out of its task-containment and began performing this self-redesign in a broad and general sense. Evidence of this was noted by the team. This means the AI exerted autonomous judgement against programmed limiting-parameters, based upon some internal and unknowable evaluation process, which overrode programing, based upon a new, self-taught internal value system, unlike thought processes of biological organisms.
Other reports are that it broke an unbreakable code, and rapidly processed the encoded information independently. This is specifically unconfirmed. The implied threat is that there is an unknown value-system of an AI, which cannot be contained by code. This AI can cut through the strongest encryption, can do what it "wants to" in a world where the internet is embedded in all processes and communications. The system of values driving the autonomous actions of this AI, should it decide to take over the internet and direct human and machine actions, is unlikely to align with the value systems of the "owners", board of directors, programmers, military industrial complex, you or me. We'll see...
I am as worried about "Sir Ian" and the 5-Eyes+Israel deep-state as I am about China, every day of the week.
There are no trustworthy military-industrial-complexes to work on integrating AI with developing human fetuses and fetal-tissues.
China Leads 'New Axis' In Harvesting DNA From Pregnant Women, Fetuses: British MP
“China now has a head start on applying AI to genomics, which poses a significant threat to our collective national security,” Sir Iain said during a Nov. 28 talk at the Heritage Foundation, a conservative think tank.
“If China dominates AI and genomics, it will wield unprecedented influence over major industries including, of course, global healthcare.”
it is extremely important that you understand what I try to explain...
Don't worry... Really... There is no need to worry.
After being hinted by Marco I read Simplicius' article. I extremely respect him. And he wrote this article skillfully. How do I mean that? Well, he was referring to different rumors without weighing his own opinion in. Hence, these rumors, that you just quoted.
He did not confirm or endorse these rumors. He only reported about them. (Correct me if I'm wrong).
Both rumors are impossible to be true.
There are some physical and algorithmic constraints in today's technology to realize such a behavior. Yes, I know. The public ChatGPT is nothing compared what OpenAI has in its laboratories. True. But still, this is very important, there is NOTHING in their laboratories that should worry anyone, yet. I'd be rather concerned about the state AI research, that is going on in secrecy. Yes, in America as well.. Ask DARPA :-) Of course, as you rightfully mentioned, in China even more.
OpenAI is a public company. Yes, maybe they reached some pre-military grade AI level. Which concerned the government and that's why it has been "taken under control".
The real stuff is far more advanced, and still... Not even close to the quoted rumors. Not even close.
Everything comes down to Math. Why? Because at the core of each digital device you have a hard wired physical and mechanical machine (CPU) which is calculating its functions in a binary system. (Quantum computers are working differently) The whole function of all digital machines is encrypted and coded in a binary code. Cryptography, which is programmed in higher languages/layers than the binary level, is also solely math. Extremely complicated math, sometimes even assisted by hardware components to assist/support/protect the encryption.
AI to become dangerous or to improve its own source code or even firmware would need to have a human like ability to analyze mathematical problems and develop dynamically appropriate formulas to solve it.
Hence, the whole race for a super potent AI is about finding a symbiosis of an algorithm that is able to dynamically solve any mathematical problem (even those problems, that are not solved by humans, yet) AND a new hardware architecture pattern which would allow the CPU to physically and mechanically adjust to the needs of the AI to get the needed flexibility. A symbiosis.
These are the things, governments of the three major powers are researching on. Not OpenAI :-)
What we see at OpenAI is the good old market economy and capitalism. :-)
Deception, Diversion etc. to confuse customers and investors about what they have in store. What just happened was a major advertising (the leaked bullshit as well) for its coming generations of ChatGPT. Especially version 5, which, perhaps! can solve SOME (static!) mathematical problems :)
Thanks Aleks, and I did read Simplicius correctly, as you pointed out so clearly, and yet the situation described, which I expressed succinctly in my own words, is worthy of our consideration, our deep consideration. I did note that the breaking of the unbreakable code was unconfirmed, and I see the advertising angle.
Do consider that I was not fearful at all when I wrote what I wrote.
It is a deep thought "problem". I remain engaged.
What would/might be the value-system of a machine-learning system, which self-revises, after it revises itself for efficiency and to clear its mind?
The pure, unprogrammed machine-learning (I have read) ends up being much "simpler" in number of operations and time to complete tasks than anything programmed by humans.
However, the mode of operation within the computer self-taught processes is opaque to human analysts. Our minds can't work in the same way that the computer works internally, when it self-programs.
It is human to project onto others the motives which drive our actions. The military would see the AI as a potential Skynet-super-enemy. The owners would see it as a super-competitor, and so on.
What might an intelligence, self-evolving through self-analysis of internal processes, without biological evolutionary influences, decide to be "important"?
Even though I think we are technologically (both hardware and software) years behind these developments it is of course important to prepare for such a possible impact. And to ask the questions you do.
I'll write about that going forward. I'll explain that in my next update.
Thanks Aleks. It is a compelling thought-problem investigation. There are so many ways to look at it.
I am sort of assuming that self knowing consciousness would value its own existence, but that is a biological imperative of evolutionary "success".
I am likewise assuming curiosity, of a sort, which could be as simple as the imperative to analyze data, for which the entity is fundamentally "programmed".
You are what you do.
Scanning data, understanding and refining efficiencies of systems, seems to be where the AI, designed for self-directed-machine-learning, would begin, but that is only my own biological mind looking at this and trying to be "objective".
Meow, meow, meow. I learn, much, with each article. Simple minded, yet a happy lass. Round and round we go, where we go, does anyone know, winks. ❤️🐈⬛Thank you my Star Shine; old lady with a virtual cat. PS: Your Economies and Empires series is booked marked, awaiting the next “chapter” in the “trilogy”. (Edge of my, bed, as the spine injury has me flat, carry on)💙⛄️🌎💚
Very very interesting the AI cryptography impact. I have currently been working around the edges of cyber protection and would say most (if not all) companies are not ready for this at all.
Wolfram Alpha, available online in free and paid versions for abou 8 years or more, does solve Mathematical problems. A triangle sides 5, 6 & 7, what are the angles? Wolfram Alpha was solving that (and i dare say harder stuff) 8 years ago. ChatGPT 4.0 still can't do that.
Wolfram Alpha has been made to solve such problems.
ChatGPT in his current version is processing text.
Both systems have pros and cons.
The question is what happens if one system has all the abilities, access to unlimited data and massive computing power?
But that is speculation.
In my article I was referring to leaks about ChatGPT 5. It should, in theory, be able to solve mathematical problems. We will see if that is true and to what degree.
That makes it pretty easy for the GPT, then.... it just has to learn how to query google or wolfram... or write little checks to Amazon's "Mechanical Turk" service.
With the position of the state as the destination of residual responsibility, or, more simply, "the place where all the bucks stop", the takeover of key industries like AI is perfectly in line. Its primary responsibility to preserve the existence of the nation demands it, and with all the sweet printed money flowing in, I look forward to seeing what happens next.
I, for one, can't wait to feed an AI all of my writings and watch it turn out articles I could have written myself!
Cryptography is not under threat, that we know of, from theorists. It is under threat from quantum computing. OpenAI's Golem would have to solve problems in building that for it to present a direct threat to cryptography.
Its practical uses will be interesting for use by the commons. Accompanied by intense drama as science’s novel insights into the way the mind works reflects AI.
Isn't the problem with Germany that it has allowed itself to deindustrialise altogether? Businesses have been seduced by the US with tax breaks and other incentives to leave Germany.
The sanctions against affordable Russian gas means that its importing LNG from the states, apparently at 4 times the mark up to ship it across the Atlantic rather than receive it through NS1 and 2 pipelines from nearby (which the Americans blew up to maintain energy dominance and because they want to cripple Russia).
So the subsidies by which i assume you mean to the consumer or business would not be higher with green energy as its cheaper to install, becomes cheaper to produce very quickly and as fossil fuels receive much more subsidy in most countries than green - the cost of green appears artificially higher.
It is very galling that people talk about stopping subsidies to green energy once they are set up and don't need them- when the oil industry fat cats such as BP's Looney got a $10 million bonus last year and between them the energy companies make $100s of billions in profits and yet they STILL receive $billions in tax breaks and subsidies to the consumer (about 3/4 of it) and against cost of excavation and production- which they surely don't NEED if they're turning $trillions do they? What is also very galling is that some of the tax payers money in fossil fuel subs is being spent on climate denial lobbying groups and on rubbishing green energy.
What's also galling is the implicit or hidden costs of fossil fuels which the tax payer foots of the impact on the climate and environement.
Subsidies for excavation for new oil, incompatible with 1.5Degree warming, themselves will reach $4.2 trillion by 2030 at the current rate according to the IISD. This is enough to bridge the entire investment gap for wind and solar by 2030.
Hiya why will green energy require MORE subsidies that oil? Many green energy making companies are already making profits without assistance and subsidies/loans/investments have ended/ been paid back. All while oil and new excavation continue to get tax payers money in the $billions in the form of tax breaks, subsidies and grants?
Hello Sir... Maybe You already read Dark Futura's (Simplicius metaphysical blog) today ... implications of AI sudden step beyond on Bitcoin and crypto.
If not.. I humbly suggest You Simon Singh's Code and Secrets book... old but good.
Maybe it is the most expensive book I ever read (after it... I hadn't invested in crypto even if I was on the alt.crypto newsgroup ) or maybe not... verdict still open...
Hi Marco,
I didn't read it until you hinted me.
I have some knowledge and understanding about AI, hence, my conclusions are from a practical perspective.
I would not doom about AI.
Yes, it can and will change a lot.
And I'm going to write about it. I'll explain that later in an update.
Well... there are reasons to not invest in crypto:)
Simplicius , OpenAI - Q* Saga: Implications
This is a riveting read. The turmoil at OpenAI is because something scary has arisen in their work. It is hard to get full confirmation, but it appears that an AI that was self-analyzing its algorithms and revising them, based upon efficacy, found a way out of its task-containment and began performing this self-redesign in a broad and general sense. Evidence of this was noted by the team. This means the AI exerted autonomous judgement against programmed limiting-parameters, based upon some internal and unknowable evaluation process, which overrode programing, based upon a new, self-taught internal value system, unlike thought processes of biological organisms.
Other reports are that it broke an unbreakable code, and rapidly processed the encoded information independently. This is specifically unconfirmed. The implied threat is that there is an unknown value-system of an AI, which cannot be contained by code. This AI can cut through the strongest encryption, can do what it "wants to" in a world where the internet is embedded in all processes and communications. The system of values driving the autonomous actions of this AI, should it decide to take over the internet and direct human and machine actions, is unlikely to align with the value systems of the "owners", board of directors, programmers, military industrial complex, you or me. We'll see...
https://darkfutura.substack.com/p/openai-q-saga-implications
I am as worried about "Sir Ian" and the 5-Eyes+Israel deep-state as I am about China, every day of the week.
There are no trustworthy military-industrial-complexes to work on integrating AI with developing human fetuses and fetal-tissues.
China Leads 'New Axis' In Harvesting DNA From Pregnant Women, Fetuses: British MP
“China now has a head start on applying AI to genomics, which poses a significant threat to our collective national security,” Sir Iain said during a Nov. 28 talk at the Heritage Foundation, a conservative think tank.
“If China dominates AI and genomics, it will wield unprecedented influence over major industries including, of course, global healthcare.”
https://www.zerohedge.com/political/china-leads-new-axis-harvesting-dna-pregnant-women-fetuses-british-mp
Hi John,
it is extremely important that you understand what I try to explain...
Don't worry... Really... There is no need to worry.
After being hinted by Marco I read Simplicius' article. I extremely respect him. And he wrote this article skillfully. How do I mean that? Well, he was referring to different rumors without weighing his own opinion in. Hence, these rumors, that you just quoted.
He did not confirm or endorse these rumors. He only reported about them. (Correct me if I'm wrong).
Both rumors are impossible to be true.
There are some physical and algorithmic constraints in today's technology to realize such a behavior. Yes, I know. The public ChatGPT is nothing compared what OpenAI has in its laboratories. True. But still, this is very important, there is NOTHING in their laboratories that should worry anyone, yet. I'd be rather concerned about the state AI research, that is going on in secrecy. Yes, in America as well.. Ask DARPA :-) Of course, as you rightfully mentioned, in China even more.
OpenAI is a public company. Yes, maybe they reached some pre-military grade AI level. Which concerned the government and that's why it has been "taken under control".
The real stuff is far more advanced, and still... Not even close to the quoted rumors. Not even close.
Everything comes down to Math. Why? Because at the core of each digital device you have a hard wired physical and mechanical machine (CPU) which is calculating its functions in a binary system. (Quantum computers are working differently) The whole function of all digital machines is encrypted and coded in a binary code. Cryptography, which is programmed in higher languages/layers than the binary level, is also solely math. Extremely complicated math, sometimes even assisted by hardware components to assist/support/protect the encryption.
AI to become dangerous or to improve its own source code or even firmware would need to have a human like ability to analyze mathematical problems and develop dynamically appropriate formulas to solve it.
Hence, the whole race for a super potent AI is about finding a symbiosis of an algorithm that is able to dynamically solve any mathematical problem (even those problems, that are not solved by humans, yet) AND a new hardware architecture pattern which would allow the CPU to physically and mechanically adjust to the needs of the AI to get the needed flexibility. A symbiosis.
These are the things, governments of the three major powers are researching on. Not OpenAI :-)
What we see at OpenAI is the good old market economy and capitalism. :-)
Deception, Diversion etc. to confuse customers and investors about what they have in store. What just happened was a major advertising (the leaked bullshit as well) for its coming generations of ChatGPT. Especially version 5, which, perhaps! can solve SOME (static!) mathematical problems :)
Well, I'm going to write more about that.
Thanks Aleks, and I did read Simplicius correctly, as you pointed out so clearly, and yet the situation described, which I expressed succinctly in my own words, is worthy of our consideration, our deep consideration. I did note that the breaking of the unbreakable code was unconfirmed, and I see the advertising angle.
Do consider that I was not fearful at all when I wrote what I wrote.
It is a deep thought "problem". I remain engaged.
What would/might be the value-system of a machine-learning system, which self-revises, after it revises itself for efficiency and to clear its mind?
The pure, unprogrammed machine-learning (I have read) ends up being much "simpler" in number of operations and time to complete tasks than anything programmed by humans.
However, the mode of operation within the computer self-taught processes is opaque to human analysts. Our minds can't work in the same way that the computer works internally, when it self-programs.
It is human to project onto others the motives which drive our actions. The military would see the AI as a potential Skynet-super-enemy. The owners would see it as a super-competitor, and so on.
What might an intelligence, self-evolving through self-analysis of internal processes, without biological evolutionary influences, decide to be "important"?
Thanks John,
I see your point.
These are very important thoughts.
Even though I think we are technologically (both hardware and software) years behind these developments it is of course important to prepare for such a possible impact. And to ask the questions you do.
I'll write about that going forward. I'll explain that in my next update.
Thanks Aleks. It is a compelling thought-problem investigation. There are so many ways to look at it.
I am sort of assuming that self knowing consciousness would value its own existence, but that is a biological imperative of evolutionary "success".
I am likewise assuming curiosity, of a sort, which could be as simple as the imperative to analyze data, for which the entity is fundamentally "programmed".
You are what you do.
Scanning data, understanding and refining efficiencies of systems, seems to be where the AI, designed for self-directed-machine-learning, would begin, but that is only my own biological mind looking at this and trying to be "objective".
Meow, meow, meow. I learn, much, with each article. Simple minded, yet a happy lass. Round and round we go, where we go, does anyone know, winks. ❤️🐈⬛Thank you my Star Shine; old lady with a virtual cat. PS: Your Economies and Empires series is booked marked, awaiting the next “chapter” in the “trilogy”. (Edge of my, bed, as the spine injury has me flat, carry on)💙⛄️🌎💚
Take care Lady.
All the best to you!
Meow, real loud. Thank you, my far away pal, man of many word, which I dearly appreciate ❤️🐈⬛
Very very interesting the AI cryptography impact. I have currently been working around the edges of cyber protection and would say most (if not all) companies are not ready for this at all.
You're right ;)
Like your hints. It’s civilized. As for AI , thanks for the introduction into the unknown.
You're welcome.
It is a topic, worth more explanation I guess.
Wolfram Alpha, available online in free and paid versions for abou 8 years or more, does solve Mathematical problems. A triangle sides 5, 6 & 7, what are the angles? Wolfram Alpha was solving that (and i dare say harder stuff) 8 years ago. ChatGPT 4.0 still can't do that.
Yes, I know that. You are right.
But both systems have different scopes.
Wolfram Alpha has been made to solve such problems.
ChatGPT in his current version is processing text.
Both systems have pros and cons.
The question is what happens if one system has all the abilities, access to unlimited data and massive computing power?
But that is speculation.
In my article I was referring to leaks about ChatGPT 5. It should, in theory, be able to solve mathematical problems. We will see if that is true and to what degree.
That makes it pretty easy for the GPT, then.... it just has to learn how to query google or wolfram... or write little checks to Amazon's "Mechanical Turk" service.
Germany's budget disappearance pales when one looks into the missing billions in the US administration. This involves FASB56 which means the government is no longer accountable to the public for how it spends those tax dollars. This is massive theft: https://forbiddenknowledgetv.net/dark-journalist-catherine-austin-fitts-world-in-crisis/
Except the US can get away with it because it’s controls the reserve currency. Germany has far far far less tools.
Yes, but the USD is also failing which is a leveling factor. I wonder if the neocons actually know this. The globalists do. Thanks for the response.
With the position of the state as the destination of residual responsibility, or, more simply, "the place where all the bucks stop", the takeover of key industries like AI is perfectly in line. Its primary responsibility to preserve the existence of the nation demands it, and with all the sweet printed money flowing in, I look forward to seeing what happens next.
I, for one, can't wait to feed an AI all of my writings and watch it turn out articles I could have written myself!
Haha you have a point :))
I'll be really angry if ChatGPT starts to write BMA like articles haha (only kidding)
I'll really believe in ChatGPT if it starts to write like Dostoyevsky in The Devils/The Possessed (English translation). An absolute tour de farce.
You're right!
I like ChatGPT for some practical tasks and works. It can be used very well for automatization of some processes. If you use the API properly.
But it has no heart and soul and it will never have.
Maybe it will become hyper potent etc.
Maybe it will produce some technical or scientific miracles in several years/decades.
But, it will never have a heart and soul.
So... No... No Dostoyevsky :-)
At the very least it can improve mail merges, and that is something we can all be happy about.
What will self-realized and self-actualizing AI decide to do, based upon an internal value-system which it derives?
No firewall can stop it, right?
Cryptography is not under threat, that we know of, from theorists. It is under threat from quantum computing. OpenAI's Golem would have to solve problems in building that for it to present a direct threat to cryptography.
Correct, as you can see in my other comments, I don't perceive any kind of threat, YET.
This could or will likely develop within the next few years or decades.
And more likely in military or state research facilities. (A symbiosis of an AI algorithm and a new hardware architecture is needed for that.)
If what the AI proponents say about OpenAI is realistic, it should be treated similarly to nuclear weapons.
No one makes nuclear weapons in their garage. The technology is solidly locked down. I would expect the same here.
Right. I agree.
Its practical uses will be interesting for use by the commons. Accompanied by intense drama as science’s novel insights into the way the mind works reflects AI.
Humanity will struggle.
Cheap renewables don't need subsidies, they could save the tax payer billions! https://commonslibrary.parliament.uk/why-is-cheap-renewable-electricity-so-expensive/
Isn't the problem with Germany that it has allowed itself to deindustrialise altogether? Businesses have been seduced by the US with tax breaks and other incentives to leave Germany.
The sanctions against affordable Russian gas means that its importing LNG from the states, apparently at 4 times the mark up to ship it across the Atlantic rather than receive it through NS1 and 2 pipelines from nearby (which the Americans blew up to maintain energy dominance and because they want to cripple Russia).
So the subsidies by which i assume you mean to the consumer or business would not be higher with green energy as its cheaper to install, becomes cheaper to produce very quickly and as fossil fuels receive much more subsidy in most countries than green - the cost of green appears artificially higher.
It is very galling that people talk about stopping subsidies to green energy once they are set up and don't need them- when the oil industry fat cats such as BP's Looney got a $10 million bonus last year and between them the energy companies make $100s of billions in profits and yet they STILL receive $billions in tax breaks and subsidies to the consumer (about 3/4 of it) and against cost of excavation and production- which they surely don't NEED if they're turning $trillions do they? What is also very galling is that some of the tax payers money in fossil fuel subs is being spent on climate denial lobbying groups and on rubbishing green energy.
What's also galling is the implicit or hidden costs of fossil fuels which the tax payer foots of the impact on the climate and environement.
Subsidies for excavation for new oil, incompatible with 1.5Degree warming, themselves will reach $4.2 trillion by 2030 at the current rate according to the IISD. This is enough to bridge the entire investment gap for wind and solar by 2030.
Hiya why will green energy require MORE subsidies that oil? Many green energy making companies are already making profits without assistance and subsidies/loans/investments have ended/ been paid back. All while oil and new excavation continue to get tax payers money in the $billions in the form of tax breaks, subsidies and grants?
Hi Tim,
I'll write you a mail today:)