AI Fiction Update: The Machines Are Coming For Our Words.

[JayMark]

OpenAI says it has trained an AI that's 'really good' at creative writing | TechCrunch

OpenAI CEO Sam Altman says that OpenAI has trained a 'new model' that's 'really good' at creative writing.

techcrunch.com

Watch out fiction writers....

The goal is to have AI generate fictions that are equal to or superior than anything a trained human can write.
I'd like to see this sample.

 

[Deleted member 266]

 

[Anemic_Vampire]

a trained human

Does this imply that there are untrained humans out there?

 

[JayMark]

I think the big tech corpos would love it the other way around because they dream of controling the multi-billion dollar entertainment industry without any need for human employees. So they are working hard at making AI capable in the arts. I think with quantum computing breaktrhoughs, we're actually closer than ever to the tipping point.

 

[Tempokai]

OpenAI says it has trained an AI that's 'really good' at creative writing | TechCrunch

OpenAI CEO Sam Altman says that OpenAI has trained a 'new model' that's 'really good' at creative writing.

techcrunch.com

Watch out fiction writers....

The goal is to have AI generate fictions that are equal to or superior than anything a trained human can write.
I'd like to see this sample.

It will be "superior than anything a trained human can write" until people realize the patterns. From which it will go downhill.

 

[JayMark]

It will be "superior than anything a trained human can write" until people realize the patterns. From which it will go downhill.

I fear it will have so much data it will simply recognize when readers are realizing the patterns and adjust.

 

[ElijahRyne]

OpenAI says it has trained an AI that's 'really good' at creative writing | TechCrunch

OpenAI CEO Sam Altman says that OpenAI has trained a 'new model' that's 'really good' at creative writing.

techcrunch.com

Watch out fiction writers....

The goal is to have AI generate fictions that are equal to or superior than anything a trained human can write.
I'd like to see this sample.

Seriously don’t think it is possible without human intervention with current tech. Maybe in 10-20+ years, but even then it will still be an average bot. Creating the most average stories with the most average prose. The only reason it is somewhat fun to mess with now is that it hallucinates quite easily, turning average to absurd on a dime.

 

[Tempokai]

I fear it will have so much data it will simply recognize when readers are realizing the patterns and adjust.

It's sure as hell is inevitable. But given that the more LLM slop will be produced, the more it will train on itself, and therefore killing the quality even further. Given the nature of the language, it's ultra unstable framework which only will work until there's something to feed into the machine so it can recreate it. To LLM to create actually something new is akin to asking a lion to speak. Sure, the lion knows the language, but it isn't engaging on the language game which is communication properly.

To make the LLM actually work long run they need to make the LLM full AI, from the fake "AI" thing they're pushing as marketing propaganda. And that's where they can't do anything about. The future ~10 years will be looking exactly the same, unless something truly revolutionary happens to make real AI (not LLM) possible.

 

[Corty]

If they train it on Hollywood scripts, I feel utterly safe.

 

[JayMark]

It's sure as hell is inevitable. But given that the more LLM slop will be produced, the more it will train on itself, and therefore killing the quality even further. Given the nature of the language, it's ultra unstable framework which only will work until there's something to feed into the machine so it can recreate it. To LLM to create actually something new is akin to asking a lion to speak. Sure, the lion knows the language, but it isn't engaging on the language game which is communication properly.

To make the LLM actually work long run they need to make the LLM full AI, from the fake "AI" thing they're pushing as marketing propaganda. And that's where they can't do anything about. The future ~10 years will be looking exactly the same, unless something truly revolutionary happens to make real AI (not LLM) possible.

Because it's not AGI, at least we don't believe it is. Chat gpt certainly isn't. Basic public use AI certainly isn't.

I think we already have AGI (Artificial General Inteligence) in the labs. If not it's very close. And we are much closer to ASI (Artificial Super Inteligence) powered by quantum computing than we collectively realize. The best we have access to on the interent (Grok 3) is a baby toy compared to what is being worked on in RD labs. But this is an issue that goes far beyond the writing space.

I'm hoping this all happens later rather than sooner.

 

[Rhaps]

Seriously don’t think it is possible without human intervention with current tech. Maybe in 10-20+ years, but even then it will still be an average bot. Creating the most average stories with the most average prose. The only reason it is somewhat fun to mess with now is that it hallucinates quite easily, turning average to absurd on a dime.

Yeah, I doubt it would be able to copy my ideas. Would it be able to make Hitler a good guy? Justify 9/11? Make a character so evil Satan would shiver and make them the ultimate good guy?

I haven't unleash 60% of my remaining power yet, in fear of being banned.

 

[Assurbanipal_II]

Yeah, I doubt it would be able to copy my ideas. Would it be able to make Hitler a good guy? Justify 9/11? Make a character so evil Satan would shiver and make them the ultimate good guy?

I haven't unleash 60% of my remaining power yet, in fear of being banned.

That being said, to this day, AI still needs to prove that it can make actual money~.

 

[JayMark]

Yeah, I doubt it would be able to copy my ideas. Would it be able to make Hitler a good guy? Justify 9/11? Make a character so evil Satan would shiver and make them the ultimate good guy?

I haven't unleash 60% of my remaining power yet, in fear of being banned.

 

[Deleted member 192215]

I notice that LLMs are nothing but Computer can Compile random words from a trained dictionary.

That is not even close to an AGI or ASI. A real Artificial Intelligence would need a artificial (manmade) consciousness, from scratch like a baby. Able to maintain itself regularly without any human intervention. It does not have an operating system to use (even linux), it only have kernel modules, nothing else. It is like a script for a programming that is able to programmed it self, and it can train it self without human intervention feeding the data.

Tbh, the post by ClosedAI looked stupid.

 

[NotaNuffian]

Does this imply that there are untrained humans out there?

*raise hand*

 

[OokamiOkuri]

Never trust anything ClosedAI says.

 

[ArlindoFrancisco]

I don't see the problem. If people like A. Writing: Well, you can do nothing about it. At the end of the day, it will be up to the reader.

 

[Erysion]

But can it write shameless smut?

 

[CharlesEBrown]

Does this imply that there are untrained humans out there?

::raises hand::

 

[Placeholder]

> I think the big tech corpos would love it the other way around because they dream of controling the multi-billion dollar entertainment industry without any need for human employees.

Nah, use ai art and word tools, fire 10% of the workers, and then implement 50%+ wage suppression on the rest.

> But can it write shameless smut?

We can train humans to do this.