Please beware of www.freebsdsoftware.org - it contains AI slop

[doul]

AI is perfectly capable of distinguishing between human-generated and AI-generated content.

 

[MG]

Are you proposing that "AI" cannot innovate?

Did it ever invent or construct something that has an approveable market value? I don't think so...

The problem also starts with a bogus term. AI? What are we talking about? Logic processing of bit sequences with a GPU projection system is the core tech. What should distinguish something that's made with it?

 

[drhowarddrfine]

Are you proposing that "AI" cannot innovate?

I would have no problems stating that as fact.

 

[cracauer@]

AI is perfectly capable of distinguishing between human-generated and AI-generated content.

Uh. Actually nobody and nothing can.

 

[LibreQuest]

Uh. Actually nobody and nothing can.

I can.

 

[atax1a]

this conversation reminds us of racing golf carts around in first gear until their engines explode

 

[MMacD]

I'm ok with 'using' LLMs for certain tasks. For example grammar and other suggestions for the author. (Despte it paying directly into the pockets of American fascism...)

But generating all the content with LLMs entirely is despicable. The information is completely unreliable and all that person did was waste tons of energy and learning exactly NOTHING while doing so.

EDIT: The website in questions here is clearly completely generated. With cross references and sections that make no sense. Even the CSS/Styling is obviously the first one that the LLM came up with. They all look like that. Whoever put his nonsense online is a complete waste of oxygen.

People need to understand that, except within very narrow domains for translation purposes, AI will remain unreliable for some very long time because it lacks sufficient world-knowledge for the things it claims to be able to do.

 

[atax1a]

as far as we can tell the LLM is either strictly inferior or at best "roughly equivalent to chance" than any other technique, and the only reason it isn't being laughed out of the room is the zillions of dollars being poured into it by the same people who brought us Bitcoin, NFTs, and the Metaverse

 

[MG]

as far as we can tell the LLM is either strictly inferior or at best "roughly equivalent to chance" than any other technique, and the only reason it isn't being laughed out of the room is the zillions of dollars being poured into it by the same people who brought us Bitcoin, NFTs, and the Metaverse

It looks to me like just further evolved search engines. I don't care what any prompt webservice with LLM can generate for me. What does it know in detail? How much answers can it provide to questions that can only have 1 valid answer? The hallucination thing is an attempt to obfuscate this when it can't tell. They don't have the information and are forced to add uncertanties.

 

[sidetone]

The site is useful as a case study. If you understand that, it's not necessarily harmful.

and start using things like e.g. www/iocaine which actively poisons AI scrapers with endless amounts of 'plausible garbage'.

From what I gather, AI in search engines will determine what is believed by most humans, right or wrong. It will say, things which haven't been discovered yet aren't possible. It will say ETs aren't on Earth yet. They don't know that for sure.

 

[MG]

The site is useful as a case study. If you understand that, it's not necessarily harmful.


From what I gather, AI in search engines will determine what is believed by most humans, right or wrong. It will say, things which haven't been discovered yet aren't possible. It will say ETs aren't on Earth yet. They don't know that for sure.

It sounds reasonable. AI as how it is now doesn't seem to be somehow constructive and able to create something useful that was never done before. Are there even things invented by an AI system that are not copied from public data but real new concepts? I don't think I ever heard of something.