Can AI (Artificial Intelligence) even compete?

[Touch The Universe]

People can also astral project supposively, and remote viewing, that seems to indicate conciousness in not tied corporeally. Also, in alien abduction, why do they flee at the name of Jesus and none other. This too is indirect evidence of a supernatural world all around us.

Ironically, the best things in life have nothing to do with things in the outer world. The greatest treasure of "being" are things of the spirit, that transcend carnality, or the flesh - pleasing it, drugs, sex, food, etc. Spiritual mindedness, meditation, prayer, faith, the word of God, or even enjoying nature, and appreciating beauty, learning/wonder, and especially music these are all "spiritual" things we engage with spritually. They speak nothing towards "show me the evidence" of this or that, hard verifiable data. They are non empirical, subjective things, that without which life would not be "good", or enjoyable, or otherwise meaningful.

[vurt]

aliens run away from the name of jesus? maybe they just don't want to go through all "that" again, after the time they took kenneth copeland?

[vurt]

actually the best things in life have everything to do with space, where you think the elements that make up your body were formed?

"we are all made of stars"
"ladies and gentlemen, we are floating in space"

[Zeisner]

"ladies and gentlemen, we are floating in space"

[jancivil]

Training data. Likely they whiped the memory and trained it again with less input data.

I recently watched a documentary on AI about something like this. They found that giving it data at all did slow it down. Wiping it wouldn't start it over and they decided to start from scratch and give it it's own free reign in learning.

WIth no data input at all it's a brick. Something that has no mind or awareness, where its reasoning (this is past LLM now) is a trick that's been shown to have hard limits (CF: the paper by Apple The illusion of Thinking) is not going to have any capability of free reign. Whether you were tripping, they were or were lying, this is way beyond the realm of possibility at this time.

Now, in the struggle to develop the LRM the Apple developers found that giving it more and more data did not make it do better at Towers of Hanoi at higher levels of complexity (more disks to stack with the rule a larger disk can't be placed on the smaller disk), what they found was it looked for workarounds when stumped (hence "the illusion of thinking"). This is, like all of whatever behavior exhibited, a function of computation. They've failed to program inductive reasoning capability.

There's a canard that appears in these discussion, 'it learns like a person does'. Nope.

[jancivil]

Recursive AI means at some level, a function being defined will apply internally to its own definition. Fractals are recursive, for example.
Programming a machine to become conscious is purely magical thinking, consciousness is not understood on the physical plane, full stop; there is no 'there', there. "Consciousness is not computational" - Roger Penrose

[BBFG#]

Training data. Likely they whiped the memory and trained it again with less input data.

I recently watched a documentary on AI about something like this. They found that giving it data at all did slow it down. Wiping it wouldn't start it over and they decided to start from scratch and give it it's own free reign in learning.

WIth no data input at all it's a brick. Something that has no mind or awareness, where its reasoning (this is past LLM now) is a trick that's been shown to have hard limits (CF: the paper by Apple The illusion of Thinking) is not going to have any capability of free reign. Whether you were tripping, they were or were lying, this is way beyond the realm of possibility at this time.

Now, in the struggle to develop the LRM the Apple developers found that giving it more and more data did not make it do better at Towers of Hanoi at higher levels of complexity (more disks to stack with the rule a larger disk can't be placed on the smaller disk), what they found was it looked for workarounds when stumped (hence "the illusion of thinking"). This is, like all of whatever behavior exhibited, a function of computation. They've failed to program inductive reasoning capability.

There's a canard that appears in these discussion, 'it learns like a person does'. Nope.

I think the "learns like a human" is just basically by "trial and error". Of course any AI must have a basic set of operating parameters to begin trying, failing, learning from. I didn't mean it as a initiate from zero.
To add to that AI experiment I think I would give it a pair of "human" type hands with that 88 note keyboard with a basic instruction of figure it out and see what happens. Of course we could get real abstract and do the same thing using a pair of six digit "hands". Like we've read in older sci-fiction stories.
The point being that if it has the data base of all known music, it still operates in that realm. If it can create, it would be with no one telling it what that is.

[zerocrossing]

I can understand you think I'm crazy because you aren't following what I'm saying.

I think you're mentally ill because I've had direct experience with a person (roommate) who spouted a lot of nearly identical stuff like this when she was having a manic episode and was off her meds. I'm no psychologist, but she's rant at me in very similar language, using very similar illogical arguments. AI wasn't a thing yet, but if it was I'm sure she's be weaving it all into her delusion as well. When it got really bad, she'd sit in front of an old TV and put it on UHF band and find a static channel and move the fine tune knob all night. She said there was a device in the TV that was monitoring her emotions.

[zerocrossing]

Recursive AI means at some level, a function being defined will apply internally to its own definition. Fractals are recursive, for example.
Programming a machine to become conscious is purely magical thinking, consciousness is not understood on the physical plane, full stop; there is no 'there', there. "Consciousness is not computational" - Roger Penrose

I don't agree with that, though what I am seeing that is a big problem with AI is that it has little or no memory or ability to reason. I think consciousness is a product of cognition and memory, and AI is seriously hampered by not having much memory, and its cognitive function is lacking any serious ability to reason. Try asking ChatGPT a question and then ask it the same question with slightly different wording, and watch how you get a different answer, even though it just answered the question moments ago. It has no memory of what just happened, and no reason to produce the same answer twice. If it produces a song, ask it to produce the same song it just produced, but with different instrumentation. It can't. It has no memory of producing a song.

Here's an example. It scoured the web for images, that I'm sure contained many millions of images of music keyboards. Pianos, synths, organs, etc. They all share a very simple repeating pattern of black and white keys. It also has access to a ton of music and information about our very definable musical system, yet when you ask it to make an image of a keyboard, it seems to just throw out a bunch of random keys that don't follow any pattern at all, not even one that's consistent in the image. That's because it doesn't understand music, or what keyboards are for. Same with text. It can chat with you all day, but ask it to create an image with text on it, and it's flummoxed. No idea what text is, or even what words are. How can something that has access to all the music information, access to most of the music, be able to make music, and yet fundamentally not understand what music is?

It did the same thing with human hands because it doesn't understand what a hand is. They fixed this, by offering additional training, but that doesn't mean that any understanding of what a hand is happened. Just that it could now do a better job of arranging pixels that represented hands better. (usually)

This is why I laugh at the idea of GI. I've yet to see a single example of AI proving that it actually understands things, and can reason as well as a toddler. What AI seems to be doing is dreaming.

[jancivil]

If you disagree that consciousness is not explained by physics, make that explanation. The first question one should ask themselves is where is it located? If you believe it's a function of the brain, show us. Be the first.

[jancivil]

I think the "learns like a human" is just basically by "trial and error". Of course any AI must have a basic set of operating parameters to begin trying, failing, learning from.

I have to recommend accessing "The Illusion of Thinking" or a detailed analysis by someone in the field. The LRM is kind of learning things that are not so dissimilar to what a person does, only it's how a lazy person approaches the problem: looks for shortcuts.

It's fascinating (and very disquieting) to me how a machine finds ways to be tricky. This is a serious level of complexity I don't know what to do with. No one that understands what this is believes there is a mind on the other end, but...

[BBFG#]

I think the "learns like a human" is just basically by "trial and error". Of course any AI must have a basic set of operating parameters to begin trying, failing, learning from.

I have to recommend accessing "The Illusion of Thinking" or a detailed analysis by someone in the field. The LRM is kind of learning things that are not so dissimilar to what a person does, only it's how a lazy person approaches the problem: looks for shortcuts.

It's fascinating (and very disquieting) to me how a machine finds ways to be tricky. This is a serious level of complexity I don't know what to do with. No one that understands what this is believes there is a mind on the other end, but...

Agree. And isn't one of the strongest appeals to using it is to escape the human trickery so many of us have experienced?

[ghettosynth]

I think the "learns like a human" is just basically by "trial and error". Of course any AI must have a basic set of operating parameters to begin trying, failing, learning from.

I have to recommend accessing "The Illusion of Thinking" or a detailed analysis by someone in the field. The LRM is kind of learning things that are not so dissimilar to what a person does, only it's how a lazy person approaches the problem: looks for shortcuts.

It's fascinating (and very disquieting) to me how a machine finds ways to be tricky. This is a serious level of complexity I don't know what to do with. No one that understands what this is believes there is a mind on the other end, but...

Agree. And isn't one of the strongest appeals to using it is to escape the human trickery so many of us have experienced?

I have to recommend not taking a single controversial paper as a result of anything. The claims of the paper itself are viewed as overstated by researchers. The actual contribution is the testing framework that allows for symbolic parameterization of problems which helps with the problems leaking into the training data, which happens.

[jancivil]

Some aspects reported are considered that way by some. I've probably looked at more than you assume. IME you're overstating the controversy; frankly, framing the specific as though general "by researchers" doesn't look disinterested.
A high level abstract packed with jargon doesn't do anything for me.

Or more to the point unless you reveal to us how well "LRM" is in fact systematically reasoning through [the given] problems at higher levels of complexity this isn't any refutation of the findings in that paper.

[jancivil]

(The gesture taking a single controversial paper as a result of anything has a lot of color and what follows isn't in itself substantative)

"Current evaluations primarily focus on established math and coding benchmarks, emphasizing final answer accuracy. However, this evaluation paradigm often suffers from contamination and does not provide insights into the reasoning traces.

In this work, we systematically investigate these gaps with the help of controllable puzzle environments that allow precise manipulation of complexity while maintaining consistent logical structures. This setup enables the analysis of not only final answers but also the internal reasoning traces, offering insights into how LRMs think. Through extensive experiments, we show that LRMs face a complete accuracy collapse beyond certain complexities. Moreover, they exhibit a counterintuitive scaling limit: their reasoning effort increases with problem complexity up to a point, then declines despite having remaining token budget.

By comparing LRMs with their standard LLM counterparts under same inference compute, we identify three performance regimes: (1) low-complexity tasks where standard models outperform LRMs, (2) medium-complexity tasks where LRMs demonstrates advantage, and (3) high-complexity tasks where both models face complete collapse. We found that LRMs have limitations in exact computation: they fail to use explicit algorithms and reason inconsistently across scales. We also investigate the reasoning traces in more depth, studying the patterns of explored solutions and analyzing the models' computational behavior, shedding light on their strengths, limitations, and raising questions about their reasoning capabilities."

https://arxiv.org/abs/2506.06941#:~:tex ... Farajtabar

Cornell University