Debt Rattle February 17 2026
- This topic has 67 replies, 16 voices, and was last updated 2 months ago by
.
-
Posts
-
@DBS
Investing in AI
It will need all the electricity on planet Earth and then some.
It’s a bottomless Pit.
When Bitcoin miners are talking about diverting electricity away from mining and putting it towards AI, you know the End is Near.
AI is like a Chinese Finger Trap
Hey, I know you’re old enough to remember them!
Cars that can be remotely deactivated video.
Just a reminder that when there’s an evacuation notice, go by foot or bike. It doesn’t matter if your car still goes if the cars to your right, left, front and back don’t.@CluelessHonky
Thank you for asking. Here it is: By sheer accident (or God’s plan) I was “in the room” at the birth of the version of artificial intelligence that was so successful that it quickly grew to be in ubiquitous world wide usage today. In fact it was something that I said to the machine intelligence project director of the DARPA funded project acronymic PLATO project at the Coordinated Science Labs at the University of Illinois around 1973 (or so. The precise date has faded from my very fully packed memory). In an hours long and very intense (friendly) discussion with the project participant/genius Dr. Paul T. Tucker I insisted that a computer based “Expert System” like PLATO would be useless until they made it good enough to interact with users just like human teachers do, through ordinary vernacular speech. I explained that “teacher machines” would be useless failures until people could interact with the computer system in the same way that people interact with people. For one major reason, languages of many types and forms are almost the only way that we know HOW to interact.
What happened as a consequence to that conversation is that Labs (and DARPA) shifted more attention onto the study of human language and speech vis a vis their ongoing experimentation to develop a THINKING machine that would outstrip human thinking. It’s easy to see the military interest in such a device. They had just used a CALCULATING device in Bletchley Park, Los Alamos and Oak Ridge to crack Germany’s uncrackable Enigma Codes, and build a working atomic bomb before anyone else could (because no one else had electronic computers.)
The unfortunate part of my chat with Paul is that human language turned out to be the key to cracking the code of human thought and communication. When the boys who wanted to build a robot brain (which they wanted to use to rule the world) shifted their attention away from all of other (mostly ‘mechanistic’) approaches to devising a machine-brain, and shifted it towards a full-court-press on deciphering the phenomenon of human speech and language itself they discovered the patterns, procedures and algorithms which enable us to talk with each other and gather and share the vital information we need from others in order for us to personally survive.
In other words, they cracked the code of human language and speech. The “product” of that breakthrough is the thing that is called an LLM, or “Large Language Model”. The silly notion that it merely “choose the next most appropriate word” is WAY less than a half-truth.
The LLM, that is to say the “model”, and complex software which interfaces with the model doesn’t just pick the best next word. It has computed WHY that word is the best next word. This newly acquired knowledge has been kept ultra-uber-top-bloody-secret ever since, because it provided to it’s owners and overwhelming advantage in the power game of acquiring control over others, The technology and the knowledge of how to use it, enables those self-appointed owners and practitioners to assemble electronic devices which simulate human intelligence pretty nearly as well as the ones in our bodies do, with the significant difference being that the AI Designers/Owners know PRECISELY how “human thinking” works. That edge enables them to easily manipulate the thoughts of other people quite a bit. In sone cases it enables clandestine control to the extent that they are unvirtuously controlling the thoughts (and thus actions) of other people, to the detriment of those people.
There are few people who know this because it has been, as I said, a jealously guarded top bloody secret since the moment they discovered it. Their problem now, it appears to me, is that the secret is out of the closet. Cat’s out of the bag.
And the bag is bigger than any of us could have imagined, or can imagine it now for that matter.
@MyParentsSaidKnow
Back in the day, when I was once an enlisted sailor, the common brag of those whose enlistments were nearly over (so called “short timers”) was something like, “I am so short that I can sit on a cigarette paper and dangle my feet”. Your question of, “what kind of bar are we talking about here”, just conjured that image like an old ghost.
“That bar is so low that I would have to pole vault to get under it.”
Let me try to say that in a different, shorter, way. The developers and owners of machine intelligence systems made a “breakthrough” in studying the structures, functions, patterns, sequences and protocols of human language that gave them an absolutely devastating and overwhelming advantage in controlling human populations by means of codifying and TECHNOLOGIZING the process human thought. So-called “AI” is the product of that breakthrough, and the havoc they are wreaking is the deliberately intended consequence.
AI isn’t magic. It is, literally AND figuratively the technology of “thinking”. It is utterly incapable of evaluating of determining the goodness or badness of the “thoughts” that are thereby produced, but they are amazingly good at simulating such a spiritual discernment between good and evil.
Here’s an example of a late comer to the party trying to explain it. He’s sort of starting n to get a neophyte’s understanding, but it’s more like a hunch. The “Real Guys” in machine language development (funded by Darpa but directed from above DARPA) are about 50 years or so beyond where this guy is at. He sounds like what early researchers sounded like 45 years ago when the light bulb in their noggins first started to light up.
If no 1 is held accountable, it is not revelation, it’s mockery. Laughing at you as they fuck your kids.
kultsommer, do like everybody, click, & open what you want to read
That’s beside the point.
I am still convicted that amount of geographic information presented is meaningless for the non-Ukrainian people.
Well, good to know that at least you can recite good part of that video that you just watched,@DBS
“The LLM, that is to say the “model”, and complex software which interfaces with the model doesn’t just pick the best next word. It has computed WHY that word is the best next word.”Neural nets do a bunch of matrix multiplications. At the beginning they map language to vectors and at the end they map vectors to language. As far as I know that’s all they do. The WHY is embedded in the weights or coefficients of the matrices involved. In a large model there are billions of those coefficients. The fact that intelligible results come out is amazing. So what is your WHY claim based on? Are you saying someone understands all those coefficients in some meaningful way or is there some other neural net technology that does something besides matrix multiplications that they aren’t telling us about? I am most interested in your answer.
To even begin to attempt to analyze humans language, for the purpose of then replicating the process with a machine that could then hopefully do the same thing, (that is, to send and receive information via language) the engineers first had to isolate and itemize what sorts of information were being conveyed in the first place. That research revealed that all of the information consisted of symbolic representations of real things, which at its most fundamental level is a “code” that can only be de-coded by the recipient if the recipient has such an association (code symbol = conceptual equivalency with the real thing being symbolized).
So the first big mystery was how does the equivalency get established in the first place? THAT research revealed that the symbol/thing equivalency is established in multiple ways, which I’ll only partially itemize here because the full list is both impractically lengthy and impractically complex to explain briefly.
Repetitive association of symbol-to-thing is the BIG one.
Repetitive association of symbol-to-thing with benevolent sources is the next biggie
Identification of sources that prove out over time to be most dependably benevolent is another.
Identification of sources that prove to be hostile, dangerous or destructive is also very bigLanguage, then, its very nature contains a symbolic representation or “model” of logical thought, and the conveyance of those symbols to another entity requires an almost incomprehensibly volume of repetitions that associations are made, modified and organized in a fashion that is sufficiently logical that it aligns with real world conditions well enough to lead to success and thus to survival.
In a nutshell, language itself is a symbolic model of a logical representation that reflects the activity of understanding and being able to survive in an inflexibly logical world. And that model is conveyed to others by sufficient repetition that the code is transferred to another individual by association.
And so is the fact that it takes enormous amounts of reiteration to establish patterns which can then be used to short-cut the process and speed things up.
In order to be successful in a REAL world the information and the consequences of using it requires that the process of organizing and exchanging these transfers of symbols mimics the logical cause and effect nature of reality. If the interchange does not succeed in mimicking reality then it leads to failures and the frequency of failures becomes associated with certain types of behaviors (lying and getting away with it often leads to short term benefit, but lying too much or getting caught at it leads to long term failure.So-called “neural networks” were FIRST envisaged as necessary, and only then was the effort made to build them with hardware and software, to mimic the “weighting” each item in nearly incalculably vast quantities of incoming data like the way human brains do it. There are very few whites and blacks in a neural network, but a nearly infinite number of shades of gray, and each with a value that might be a different value the next time that data point runs through the system.
The grossly over-simplified operative point is this: COPY the way language works and you copy the way a mind thinks.
The caveat is this: without a conscious awareness to evaluate and judge the outcomes provided by such a mind (whether organic or artificially built by humans, or their equivalent) there is simply no way to keep it aligned with REAL reality, and thus its ultimate failure is 100% certain. Since machines have no consciousness machine “intelligence” is incapable of such discernment, so giving machine intelligence “agency” is not just an illogical thing to do, it’s an attempt to re-write the immutable law of the Universe without the actual ability to do so. Simulating thought is a piece of cake. Creating the consciousness capable of evaluating is simply not. ONLY God can do that.
6 feet apart or alone with your phone;
Junk on a screen or junk in your arm;
Dope for the dopes and sweets for the drone:
The cattle are lowing. All’s well on the farm.They’re leading us to think the end is near.
Just isolate and you can dodge the fear.AI is so smart!
Yes, AI is really smart!
Supper dupper smart!
Amazingly smart!Let me count the ways!
A week ago, I looked at humidifiers.
Since then, all my ads are still for dehumidifiers!
Like in artifical de-smart!When I type in the word “gold”, it autocorrects to “good”!
Good, good, good, good, good, good, good, amazing AI brilliance!
Now, you know why I talk about “good” so much, when everybody else is talking about gold!Related smarts, when I type the word ;”hold”, it automatically corrects to “hood”!
Hood, hood, hood, hood, hood, more amazing AI brilliance!Artifical smarts!
I think the shortest way of saying all that is that Artificial Intelligence was not developed by invention, it was developed by copying nature, and the nature that needed to be copied was human language. They had been trying to build artificial thinking machines for ages, with neither success nor progress because they didn’t actually know what operations such a machine would actually have to perform. What they discovered in human language turned out to be what those operations had to be, and once they knew that then they knew how to build a machine to do each of those specific things. Voila! A “thinking” machine.
It’s snowing here, and it’s FORTY-FOUR degrees F outside!!!!!
WTF?but what temperature is it up there in the clouds?
4 Ways For US To Go Broke:
Actually there is a fifth way for the US to go broke, which interestingly isn’t mentioned at all!
It is the path the US is currently on!
Naturally, they don’t want you to know this, even though they have been publically telling you!
They are counting upon you dismissing all of their public statements, as lies!
Lips moving, therefore lying!What they are doing is financial repression.
Suppress short term “government” interest rates.
Move government borrowing to short term Treasury Notes/Bonds.
Under state inflation.That is exactly what they did after WW2.
That is exactly what Biden did too.
But of course you know I am lying to you, right?
The whole AI thing is a little bit like airplanes. Basically they copied birds and in that process discovered that the lift produced by air flowing around a wing had more to do with it than feathers or flapping.
In AI the big discovery was that language was the accumulated consequence individuals conveying information to one another in order to obtain the information they needed to survive but could not possibly get enough of all by themselves without helper-input from others. Language is like an interactive working “map” of thinking. Language IS the model that the Large Language Model is a model of.
tdk:
Well, you know cloud computers need super cold temperatures to ccompute!
It is snowing here in Toronto too, on the first day the maple sap ran this spring!
@DBS
You didn’t answer my question. Can the WHY be extracted from the billion coefficients in an LLM or is there something going on beyond coefficients and matrix multiplication. This is a simple question and doesn’t require an exposition of life, the universe and everything. You claimed that LLMs compute a WHY in some secret way and I’m asking you to explain how that works.is there something going on beyond coefficients and matrix multiplication [?]
The short answer is yes, and the long answer would take more pages of text than I have time to write or you have time to read. What I’m attempting to convey in a reasonable number of words is that things like coefficients and matrix multiplications are merely the calculations. Calculations themselves do not mimic human thought. Language, on the other hand, does. Compute is just compute. The only reason compute is needed in such huge quantities is that language involves incomprehensibly vast amounts of differently “weighted” items of sensory inputs which are stored, manipulated, retrieved and used using our organic “AI system”. The deep study of THAT system, the “human language system” is what revealed how the system worked. It’s “architecture”.
I cannot replicate for you the tens of millions of man hours, computer time and literally trillions of dollars that went into that project, but I can tell you that so-called “AI” is the electronic copying of the human language system, which rather miraculously enables two or more individuals to bridge the gap between individuals, using coded symbols that are learned through repetition, association, and procedural protocols that materialistic sciencey guys just aren’t equipped to grok.
That’s why they need weirdo observers like me to point out to them what they need to geek out on.
So, yes, there is some secret way that LLM’s compute the why a particular word is chosen, and I just explained it to you pretty damned well considering I only used a few hundred words to do it.
Over the coming days and weeks I’ll be explaining it in more detail, but I warn you advance that it ain’t gonna be very much about math, electronics or software. Any deep pocket technician can do that stuff, and there aren’t any real secrets involved.
The secret shit is the part that I’m telling you now. Human language is WAY deeper and more significant than was previously known, and when the bad guys got their hands on those secrets they kept it to themselves for as long as they possibly could, and took over the world with the power that it gave them.
It’s going to take some time to lay it all out, and I don’t mind questions, but don’t be too surprised if I take umbrage at unkindness and disrespect.
- You must be logged in to reply to this topic.
Sorry, the comment form is closed at this time.