• Welcome to the Fantasy Writing Forums. Register Now to join us!

Got told AI did a better job than me.

A. E. Lowan

Forum Mom
Leadership
As should shock absolutely no one, I'm falling on the more complicated side of the fence on the AI question. I don't think AI can independently invent anything. Extrapolation can look like invention, and I know because I do it on the daily, but it takes a soul, a mind squishy in both thought and texture, to create. I do think that this is a time of incredible opportunity to experiment and to freely share with each other the amazing things we, humans, can create together.

Note: Freely.

I do think that churning out things in minutes with AI and then attempting to sell it as one's own word is deeply dishonest as well as dishonorable. We all know where the information is coming from, no one should suffer from any illusions that generative AI is anything more than a chimera of the human experience.
 

TheKillerBs

Maester
Not sure what mood means, but i made a statement of fact.

Doubtful is fine, but that will be the last hold out. An Ai does not have to know why, it just needs enough trial and error, which it can do at a much faster rate—assuming of course, it is real ai.
And therein lies the issue. It is not real AI. It is a pattern-detection and continuation machine. It scans an input, finds a pattern, and then spits out what should statistically follow, or generates a statistically-likely pattern that fits a set of criteria that was asked of it. That's why it will never be better than the best data it's given.
 

pmmg

Myth Weaver
Okay, and it will never improve? It makes no sense to bet on things remaining as they are.

What a computer can do that you cant, and try a million samples at once and measure the response. It wont take it long to see what sells and what doesn't. Or what gets rave reviews and what doesn't. It just needs to be set loose, and start collecting the data.

The LLM we have today is not what we will have tomorrow. All the best we have in the Computer Science industry are working on it. A collective effort to advance. Writing is a solitary effort. We wont keep up.

I think AI may get prevented, cause those in charge will want to curtail to their biases, but, it remains to be seen.
 

TheKillerBs

Maester
Okay, and it will never improve? It makes no sense to bet on things remaining as they are.

What a computer can do that you cant, and try a million samples at once and measure the response. It wont take it long to see what sells and what doesn't. Or what gets rave reviews and what doesn't. It just needs to be set loose, and start collecting the data.

The LLM we have today is not what we will have tomorrow. All the best we have in the Computer Science industry are working on it. A collective effort to advance. Writing is a solitary effort. We wont keep up.

I think AI may get prevented, cause those in charge will want to curtail to their biases, but, it remains to be seen.
A river can't run higher than its source. It is statistically impossible for something derived from a set of data to go outside of the extremes of that data. You can, at best, get it to remove the worst of the training data and refine it so that the better input gets prioritised. But you cannot make it better than its input. It is mathematically impossible and at the end of the day, these things work using mathematics, not language.
 

A. E. Lowan

Forum Mom
Leadership
Okay, and it will never improve? It makes no sense to bet on things remaining as they are.

What a computer can do that you cant, and try a million samples at once and measure the response. It wont take it long to see what sells and what doesn't. Or what gets rave reviews and what doesn't. It just needs to be set loose, and start collecting the data.

The LLM we have today is not what we will have tomorrow. All the best we have in the Computer Science industry are working on it. A collective effort to advance. Writing is a solitary effort. We wont keep up.

I think AI may get prevented, cause those in charge will want to curtail to their biases, but, it remains to be seen.
Of course AI will improve. It's able to refine itself so fast you can see it reflected all across Pinterest. This week is stunning. Next week it will be mind-blowing. But what it is not is human. It lacks a connection to the Collective Unconscious and that intimate familiarity with what it means to be connected with other humans, and thus to write about the human experience. What it can churn out is really beautiful, and I know I get a lot of benefit from the sparks AI works can fire off in my brain, but the work itself is soulless. All flash, no pan. All style but no substance or depth.
 

pmmg

Myth Weaver
Well, if y'all say so, but time will tell.

I think AI will very likely become too hard to tell apart from the real thing. And a day is coming when it will produce things...art... that will win great appeal as great art that touches the soul.

A written work may be harder than, say, a portrait, but given enough time, it will navigate that too. The real challenge for AI is can it be real AI. If it is, it can rise higher than the river.

I don't actually care much what AI can do. I care what it means to you. It makes you not matter. I want us to matter. Bring me your heart and soul, dont bring me regurgitated AI that you doctored up to make pretty. Be a writer. Be what you were made to be.

If you are using AI, you are not a writer.
 

Miles Lacey

Archmage
To me AI is a tool, nothing more. Like any tool it can be used for good or bad. A hammer can be used to hammer nails or pull them out. It's also good for bashing things with, including people.

I did write a 15,000 word story using AI not so long ago. It was to see what it came up with. It was truly AI slop but the primary purpose was to see its flaws and limitations and whether or not it could enhance my writing. Both my slop and the AI generated slop was about the same level of terrible so there were no gains to be had by using it.

The real problem with AI isn't that people use it. It's that people use it without really knowing, understanding or comprehending what has been written. In fiction, it's annoying. In non-fiction it can have serious real world implications. Fake news, fake history, fake science... you get the idea. If used for doing exams at university level it could have potentially deadly consequences in many occupations. After all, would you want to be operated on by a doctor who did their exams and papers using ChatGPT?

For writers who go through publishers AI is not a real threat. Publishers are very effective gatekeepers to prevent any sort of slop from getting through. People who publish online mostly produce slop and it's that slop AI draws upon. And, lets be honest here, AI will not be producing anything even remotely good quality for a while yet. When the quality of what appears online improves then we can worry.

For now, what we should really be concerned about is the growing numbers of people who have low literacy skills and those who can't afford to buy the books that are being published. That is a greater threat to your livelihoods.
 
Top