Leo, You are Living Under a Rock When It Comes to AI

[something_else]

20 hours ago, Leo Gura said:

If what we had was truly intelligent and creative, then we wouldn't be calling it AI slop, we would be calling it AI genius, and people would be paying extra to view it.

Why isn't AI producing grand works of art that surpass human-made work? It's not even close.

Where is the AI Shakespeare?

Show me one book written by AI that I would want to read.

It's simple really: If AI was so great, then you could charge extra for AI work. But the opposite is the case. AI work is often nearly worthless.

If AI work so was great, everyone would be slapping "Made with AI" on their products and services. People would even be trying to pass off human work as made by AI.

It doesn't need to be able to do these things to have a significant impact on the world, though. The vast majority of the jobs in the world don't require much creativity.

You're comparing AI to the absolute peak of humanity, which of course it cannot compete with. If it could it would be ASI. But when comparing with the average human it's very capable in many fields.

[kavaris]

Oh plz no Ai @ Mcdonalds, Ahhhhhhh They are the only company that still operates somewhat reasonably (tho the toilets are, like, they got poop in em) Like they got that siq machine w/ the Mcdonalds OS on it, that imalways tryina see if theres a way to hack that bad boy when they arent lookin, so i can order myself like 17 burgers on the fly, Like weneed that Mcdonalds unscathed, atleast til im dead in like 10-20 years ya know. I cant eat there that often, but id def be there alot, if it was free, as thatd be auwesome.

Edited yesterday at 05:43 PM by kavaris

[JoshB]

On 8/24/2026 at 1:26 PM, Leo Gura said:

If what we had was truly intelligent and creative, then we wouldn't be calling it AI slop, we would be calling it AI genius, and people would be paying extra to view it.

Why isn't AI producing grand works of art that surpass human-made work? It's not even close.

Where is the AI Shakespeare?

Show me one book written by AI that I would want to read.

It's simple really: If AI was so great, then you could charge extra for AI work. But the opposite is the case. AI work is often nearly worthless.

If AI work so was great, everyone would be slapping "Made with AI" on their products and services. People would even be trying to pass off human work as made by AI.

Not sure if this is a correction of my statement but we are on the same page. All though I will say I've seen numerous amazing examples of AI doing amazing shit which wasn't possible 5 to 10 years ago. It's really understanding its limits and how developed who's using it is. 

[Leo Gura]

58 minutes ago, JoshB said:

Not sure if this is a correction of my statement but we are on the same page. All though I will say I've seen numerous amazing examples of AI doing amazing shit which wasn't possible 5 to 10 years ago. It's really understanding its limits and how developed who's using it is. 

Some humans are using AI to make good stuff. But its those humans who are making it good.

[Jirh]

On 25/08/2026 at 6:43 PM, something_else said:

The vast majority of the jobs in the world don't require much creativity.

My friend has the least creative job in the world: transcribing text from simple hand-written pages.

And even though it's good, AI still can't replace him. It might even reach a 99% accuracy rate, but he'd still have to double-check afterwards, which for him takes more time than actually transcribing.

[CosmicExplorer]

55 minutes ago, Jirh said:

And even though it's good, AI still can't replace him. It might even reach a 99% accuracy rate, but he'd still have to double-check afterwards, which for him takes more time than actually transcribing.

That's the main reason why I think current AI will not get to AGI, nor replace white-collar jobs, reliability. Humans are much more reliable than even the current best models. Even if current LLM based AI reaches something that can be called AGI for average scores, the reliability will lag behind. It's extremely hard for this architecture to reach 99% reliability score, even on simpler tasks, and humans are usually more reliable than 99%

In the Agents Last Exam benchmark, current AI has a 10.5% "pass rate"/reliability on real-world, economically valuable tasks. Not even close to enough.

[kavaris]

Like its the misunderstanding itself, in literal protocol fashion, and we are saying "wow, look how this thing misunderstands", Like, that wouldnt have been acceptable in the years prior, As they wouldve said "This thing cant manage the tribe or curate the farm". Like its not sharing a common relationship w/ what would be most necessary. Its only like this very thin veile at the surface of "human confusion and communication misses"

[Jodistrict]

I recently bought a new vacuum cleaner and one of the parts wasn’t clicking in correctly.  I used the AI and it was just leading me in circles.  I even uploaded several pictures of the part at different angles and that didn’t work.  I then sat down and thought hard for a few minutes and figured it out myself.  That and other like experiences made me realize that the current large language models do not having reasoning ability.  They are basically fancy pattern matchers.  They can appear to solve problems if the problem and solutions have already been recorded in the data they have access to.  But the problem solving is an illusion.  The AI is just pattern matching previous solutions.

Edited 2 hours ago by Jodistrict

[kavaris]

33 minutes ago, Jodistrict said:

The AI is just pattern matching previous solutions.

Yeah. The whole pattern thing has been blown out of the head of proportion, its like this thing where these anax patterns, being that theres infinitly-many patterns, are taking priority over basic foundations, when they should be becoming aware to the fact that its a one-of-many, too many, and to name them all is just like naming all the factors in atoms, the universe, and the observer, and things like this, its like, dang... dang, thats too many things. Like iuno what youd call that, when its simply too many patterns, with too much focus, like an "over-something" ( i cant think of wat it is atm )

Edited 2 hours ago by kavaris

[kavaris]

Like i feel like Actualized is also a way to resist the unresistable, like its both how to see into the bs mechanisms — to not flutter on into the big business machine, and its also how to become aware to the primary machine that enters into the experienced ovation of self, discipline and wisdom, like its the means of becoming something that touches all of these things at once. Which is so cool, for just the basic human pattern mechanism thing, whatever we are to decide.

[Schizophonia]

Leo has built himself the persona of a serious conscious businessman in a white shirt.

If AI arrives and enables universal basic income he will no longer be able to gain value through his social status and he knows it; so he's using a coping mechanism convincing himself that nothing will happen.

 

What is he ultimately going to manifest? 👀

Edited 1 hour ago by Schizophonia

[Schizophonia]

When hookers' clients realize that universal basic income is going to send prices skyrocketing since the women won't have to do that work anymore they’ll become AI-skeptics too. 😸

Edited 1 hour ago by Schizophonia

[Jirh]

1 hour ago, Jodistrict said:

I recently bought a new vacuum cleaner and one of the parts wasn’t clicking in correctly.  I used the AI and it was just leading me in circles.  I even uploaded several pictures of the part at different angles and that didn’t work.  I then sat down and thought hard for a few minutes and figured it out myself.  That and other like experiences made me realize that the current large language models do not having reasoning ability.  They are basically fancy pattern matchers.  They can appear to solve problems if the problem and solutions have already been recorded in the data they have access to.  But the problem solving is an illusion.  The AI is just pattern matching previous solutions.

AI chatbots are language models. They have some form of reasoning, which is more inductive and pattern-based. It's top-down reasoning, but it's still valid reasoning nonetheless, even though preliminary and incomplete. 

The problem is how to train these models to reason properly bottom-up without them becoming dogmatic. It's a paradox. Because they're only able to learn through a reward system. So if you reward them for valid reasoning, they will take that reasoning to heart and become dogmatic about it. Then if another valid, yet contradictory reasoning comes up, it will not know what to do. So it will probably fallback to pattern matching in order to determine which one is more appropriate. But then again, it will generate the answer from previous patterns, so it's not going to be perfect.

This is because it lacks meta-reasoning, aka consciousness. Every reward and optimization will drive the model further down a fixed objective functionality, which becomes narrower the more it is trained.

AI cannot escape this paradox. At least that's my current understanding.