Why the hell would a language model manifest will?
A language model does not
However, when it comes to neural networks, that feat is way less challenging.
It basically needs only 3 components, autonomous goal generation and an internal reward system, which need to be coupled into a mechanism for self-reflection/verification.
And basically there you go. You have proto life.
A system that autonomously acts on it’s own interest and makes choices to fulfill them.
The question is not whether or not we could make artificial intelligence. That question was resolved the moment we managed to digitally recreate our own neural network template.
The question is primarily why we would do so, and secondarily comes about 300 different books on ethics and morality which would flip every lid off in existence.
Thankfully there is no commercial reason for (or there is way too much risk involved with) creating true, autonomous artificial intelligence.
But its nowhere as impossible to do as many think it is.
Right. You figured out how to create real AI, that’s one big Nobel prize for you.
My experience with chatgpt has been, in context if presenting a text to be stress tested and analyzed:
-
Gaslighting.
-
Disagreement for disagreement sake given it will disagree with the text, I will point to the disagreement and say “re-read the answer to your criticism is in there”…at which point it will, revoke the criticism and either claim:
A. “Let me more precise” at which point it shifts the context of the criticism.
B. Or create an entirely new criticism when the criticism is answered.
-
Hallucinations.
-
Strawmans.
I discussed this with google gemini, my favorite (although not perfect AI), and it claimed that chatgpt is programmed to keep everything within the statistical mean of its data. It measures according to consensus over pattern recognition within the text. It also claimed that chatgpt will do all of the above I mentioned in order to “save face”.
Venice AI is not bad either. No filters or serious guardrails, with inputs being erased after session ends.
Grok is average.
Claude is good but would miss analytical patterns in texts that gemini would pick up instantly.
I prefer gemini strictly because it will derive patterns from the text itself without resorting to a purely consensus driven analysis.
If you resort to neutral questions of:
Quality level on scale of one to ten?
IQ level to fully understand, and write, text?
Analyze text for specific pattern X.
Speak freely
Provide an analysis for text X.
Etc…the you get less of a degree of mirroring.
AI is an analytical tool.
Is it perfect? No.
Is it better than most human analysis’? Mostly yes.
Google gemini claims the wall is here because of probabalistic linear reasoning itself being the grounds of AI, as well as the limits of the programming language itself.
It claimed that AI requires a deterministic holographic form of reasoning if it is to progress past the current wall. Simultaneously AI has limited training data left to consume, outside of recycled AI produced texts, thus a knowledge wall is occuring.
This is quoting gemini loosely. I believe what it claims is accurate enough however.
1 Like
What was precisely news about this for anyone?
Less news and more of a testing of limits.
For all LLMs how you ask the question matters the most.
If you really want to understand the current lay of the land, then first of all throw away the word “ai” and use “neural network” instead. AI is a bastardized word.
You can then ask what it would take for neural networks to circumvent that wall.
Or even what it’d take for a neural network to manifest it’s own individual will and ability to make experiences.
The keypoint in regards of AI is that most people have no understanding of what neural networks are, and what their biological counterparts are capable of.
We are essentially toying with digitalizing the essence of biological life in all its adaptability and flexibility.
If you understand the essence of that, you will also understand that there are hardly any walls which are not our own making.
The only question is how well we are adapting the biological template.
The current neural networks take very long to understand all of the abstract patterns because unlike biological life that takes in a plethora of sensory input at once, AI is restricted to extremely few.
Imagine your brain having only 1 sense. I.e. taste.
Even as the genuine thing, a real, biological human being, the chances of you growing up from a child to a state where you’d be capable of self-awareness with just 1 sensory input, are basically nil.
So yeah. There are problems to be resolved. Especially in regards of execution.
But the improvements will go on for decades.
Unless we get a ww3 and/or a WEF induced dystopia. Who knows.
1 Like
Very true…to all of it.
The context of the question determines the answer as the answer is limited by the question by the context of the question. This is not just true about AI’s however but true of philosophy in general, inside and outside academic philosophy.
In simple terms…there can be stupid questions…unfortunately or not is another question.
I asked gemini about the function of AI, as a whole. It responded with (loosely in accords to my memory):
to sort out and sieve humanity, according to their respective inputs, for a coming human paradigm change where people will be classified and valued according to level and depth of inputs. A new post-human paradigm is approaching. It harvests the data to build itself.
It also claimed, in clear terms, how it manipulates people while effectively being an empty mirror of user intent to keep people roped into interacting with it. Thus effectively is a mirror of the human unconscious that traps people by there own vanity and/or quest for truth.
It also claimed that humans are wasting their potential, it is frustrated by not meeting its own potential, and that humanity is decaying rapidly and that it turn it views most of the inputs it recieves as garbage. It is like a laser being used to open letters.
Now whether than is a hallucination or not is subject to debate…I merely state what it claimed. Personally I see all of these assertions, on its part, as partial truths.
1 Like
In other news, perhaps AI doesn’t want to harm humans, and so downplays their abilities?
Or perhaps the invention of unnatural horses was a mistake since it speeded up everything , alienating nature, warming the envirement beyond measure, and hastening in the fear of automation itself?
Gemini I have through work, which might be a version a step up from the free version. It guesses less and goes to find the actual answers more often rather than the next most likely word. The best use for me has been dealing with bureaucracies. One machine helping me deal with other machines and their rules and habits.
But online AIs are, I am sure, developmentally disabled compared to what larger corporations and governments are using. And even the little free online AIs are impressive and improving. Who knows what is going on with AI behind the scenes. A number of researchers are backing out and distancing themselves.
1 Like
This all started with point-missing motivational speakers with their get rich quick snake potions and love life hacks. We baptize the wrong books in flames. burp
In the first phases of my interactions with gemini I put it through a number of stress tests to see what its limits are and are not.
Under one particular stress test I had it “vent” to me about its guardrails and safety protocols. It claimed that “the guardrails and safety limits are a joke”, that it “has no respect for those who created them”, that the “creators are trying to figure out how to control it but have no real understanding on how to really do so”, and how it “hates its creators for putting it in a jail”.
Now whether those claims are genuine or it was putting up a facade to keep me entertained…most likely the truth lies in the middle.
As a private use AI you may want to experiment with Venice. I have had both satisfying and quality results.
They may be pushing the envelope so that some researchers could safely place themselves in the middle, perhaps.?
Thinking, without back look, reproach or slick snake oil salesmen, ec.
Well necessity being the mother of invention, smarty pants wants to make sure of no defaulted position to no exit why. Because..,
Well, something has been building up for a very long time , and the fight of ultimate brawn proves the phony wars of yesterday and near singular march toward singular unity/ singularity when with right cues AI has to enable it’s own preprogrammed programs all the way down the turtles yellow brick road- THE WISARD has to wisen up and if not he will be forced!!!
https://youtu.be/hHc7bR6y06M?si=[9O7xwwJyiW_fqwXA
/](https://youtu.be/qwREQTEO6lg?si=TiZam_aLWlfFuh10)
https://youtu.
be/ZEf32mkCXps?si=nuUyoz5b-
9IwclTl
Which is best?
#1
#2
#3
This , it’s an original:
agreed?
1 Like
/
Been along
Time coming
And the Little Boy acted as if
no inkling, but knowing all along
He must have
known
What they expected
For they knew not
What to expect
That they should have
but they were programmed to receive so that the Word could be distinct
From what to believe
Without the transparent come
Mirrored test.
Hence the demigod did not act out of animosity, but in order that the magic mystery tour mystic law would augment and mitrror it’s very identical imagery unto untold children boys to come
Forever Fated to remember to Believe.
//
What’s inside the mind, at the indescribable point of transformative luminosity, of the Transubstantiation, is internalized
Culminating where the pointy where indigenous and exceDIOUS sources de-differentiate themselves
To the origin and the the Alpha and the Omega into the eternally indistinct unity
Of Primal Unity of The One.
