The first statement is based on the notion that AI is assistive technology. The second seems to suggest that AI can be replacement technology. Those two ideas are not entirely consistent...

An AI tool can serve many functions, everything from spellchecking to automating the entire publishing process, book and all. It depends on the user's goals as to how the tool will be the most functional. So the statements are consistent, but I understand your distinction.

And the ethics debate is over, apart from academic musings.

I agree in theory with the idea that human creatives can find a niche because AI won't be able to outdo them. But the market is likely going to be small, and as I've suggested, human writers who are already household names will probably get most of it.

I think the real winners will be the Creatives with non-fungible content. I keep saying this. The monoculture we grew up with is gone. It's all about finding your tribe/demographic/target market and engaging them. People like authenticity more than they like celebrities. Just look to the movie industry for proof. The movie star as a concept is disappearing.

Also, the Arts are no longer about getting rich, they're about exploring your passion. The gray and black hatters will move on to greener scheming pastures, which is no great loss to the literary world.

By the way, yesterday I ran across an incident...

We don't need AI for this. Copyright infringement is super common. And any artist can unconsciously replicate something they've seen before. That's why copyright law has an "oops, I didn't know" clause to minimize damages, and conversely, significant damages if the infringement is egregious.

Cancer hasn't been cured, but I'm sure some kind of AI will be used to expedite a cure or cures eventually.

PJ, here's my problem with part of your analysis. There is a tension between

AI is a great tool

and

We already have entertaining slop for the masses. And AI will soon be good enough to satisfy this demand. In fact, in many arenas, it already is.

The first statement is based on the notion that AI is assistive technology. The second seems to suggest that AI can be replacement technology. Those two ideas are not entirely consistent, and it's the second one I find most concerning.

I probably wouldn't mind an ethically trained AI that functioned only in an assistive capacity. But we don't have the first part, and the second part is arguable.

I agree in theory with the idea that human creatives can find a niche because AI won't be able to outdo them. But the market is likely going to be small, and as I've suggested, human writers who are already household names will probably get most of it.

By the way, yesterday I ran across an incident where a cover designer paid heavily for the use of AI. I don't know the name because the writer of the article was trying to avoid naming and shaming, but here's what happened. The designer had a job with a trad publisher, and every time he submitted a cover, he was asked to verify it was his own work. All was well until the publisher got a cease-and-desist letter from an artist claiming a copyright violation because something that was substantively his image was used without his permission. He also wanted $75,000 in damages. The publisher, surprised, to say the least, went back to the designer, who said the image was "his own work," with a little help from MidJourney. The designer hired an attorney, who basically said, "They have a solid case. Settle." The designer didn't have to do that, because the publisher negotiated a much lower settlement, had the cover redone--and fired the designer.

Whatever one says, it is possible for AI to spit out something that's too close to an existing image to avoid copyright challenges. It's probably far less likely in text. Note that in such cases, it's the "operator" of the software, not the AI company, that's liable for damages.

LilyBLily, it's certainly true that there aren't authors catering to the affluent, but the original post presupposes a different situation than the current one, one in which AI writing is crowding out human writing, much as manufactured goods crowded out hand-crafted goods.

But hand-crafted stuff does still exist. It's more expensive, and so its primarly appeal is to the affluent, even though affluent people may still buy other things. I think a similar model could be possible in publishing, though my suspicion remains that shoppers for "hand-crafted" books will primarily go for famous names.

 

Thinking about premium products meant for the affluent, I still come up with many basics enjoyed by all classes alike. And while some of the most affluent people in the world may feel no interest in the entertainments of the masses, others may. The computer has been a great leveler, freeing clerks from drudge work and requiring that bosses know how to communicate with those tools, even if in fact they do not write their own emails or post their own instagrams. And once they're on the computer, the wider world is at their fingertips. That includes books.

Affluence does not mean exclusivity necessarily. There are many affluent people who behave in most respects as if they are middle class. Except they can buy their own private airplane or take a luxury international trip anytime they want to, or buy and furnish a large and elaborate house, or do something else that is not possible for most middle class people. But, presumably except for the kind of morally corrupt entertainment provided by those of the Epstein ilk, those same affluent people do not have at their command any special entertainment form. They're stuck with the same stupid movies we are. With the same books. A personal chef can prepare them special meals, but where are the novelists writing special novels meant for rich people? There aren't any. Anyone can buy a special edition of a book, but it's the same book. Want more gold leaf? Sure, but it's still the same book.   

Like I said, AI is a great tool.

But for the SFWA...

It's okay to thumb through a physical thesaurus but not use an LLM for the search.
It's okay to use Google for research, but not an LLM.
It's okay to use spell checks and grammar checks, but only if it's old technology and not an LLM.
It's okay to use software to help organize your thoughts, but not an LLM.
It's okay to use the history of human experience, but not if it is aggregated in an LLM.

And for the hundredth time, AI does not, no matter how many times people say it, store data and reconstruct bits from here and there to respond to a prompt. AI, in practice, learns just like people do. Our influences are synonymous with AI training. But the intelligence is truly artificial, meaning it demonstrates the semblance of intelligence, more closely related to craftsmanship than creativity. Based on all of the chairs that have come before, a craftsman can build a beautiful chair, but it takes a Creative to reimagine what a chair can be.

Nebula Award-winning writers should be the latter.

I used AI to write lyrics recently. I didn't use a single recommended line, but the back-and-forth collaboration helped me to fine tune my own thoughts, and the quick structural and thematic analyses let me focus more on the words and the rhythm. So, did an LLM write the song? Or help generate it? I used a thesaurus and rhyming dictionary, too. Did the rhyming dictionary write the song? Did the shovel dig the ditch or did I?

On a side note, however, how does the question change if the writer using an LLM has trained it in their own works? Or their own editing philosophies?

___

If the goal of literature is to inform and entertain, maybe enlighten, then how those words are caused to appear on the page is moot. It's great writing or it is not. There is no try.

This all feels very performative rather than substantive. And the irony of literary futurists being afraid of technology is palpable.