The AI author panic has overtaken writers’ conferences and podcasts since the start of the year. Endless experiments with ChatGPT and the current generation of Generative AI has the author community spooked.
Author Tiffany Yates Martin even asked ChatGPT to explain itself as a potential threat to authors.
So what’s the verdict?
It is Artificial?
Very. ChatGPT cranks out the most soullessly bland text imaginable. Indexed to the hilt for style and readability scores, the text it produces is deadly dull.
Is it Intelligent?
No. ChatGPT and its ilk are highly complex, rule-based systems, capable of extrapolation and interpolation on a massive and blindingly fast scale. That isn’t the same as intelligence.
How does it work?
Right now, all the generative AI’s are sophisticated Language Models. Or, as critic Jeff Jarvis calls them, Word Re-arrangers. They scrape vast amounts of text from the Internet, cross-reference it, tabulate it, and take a reasonable guess at context. They then apply the language model to produce new text.
Are they creative?
None of them are capable of taking the imaginative leap to true creativity. They can’t invent anything new that doesn’t already exist on the accumulation of found-facts scraped from the Internet. Occasionally they add 2+2 and make 5. Or 3. Or zero. Independent fact-checkers are calling out inaccuracies in AI content daily.
The majority of attempts at fiction or poetry are lamentable.
Can you tell the difference between AI and human written content?
Not always. A lot of human written content is also terrible.
Should I be worried that the AI will replace me as a writer?
Probably. But not yet.
The current generations of AI’s can’t imagine or think. But then a lot of human writers who ‘write to market‘ appear not to, either. If you consider there are no new plots, only new twists, it is possible within a few years that AI might generate formulaic, mass-market genre fiction that is indistinguishable from that produced by humans.
At that point, AI might crank out genre fiction in the hundreds of thousands and crowd human authors out of those very markets. We won’t be replaced, we’ll just never be found or read. Especially when AI can also generate professional-quality cover art and synopses. Without spelling or grammar errors.
Can we identify and label AI-generated content?
Not yet. The AI-detectors are equally lousy and label both AI and human-written content as AI (‘fake’, ‘plagiarised,’ ‘manufactured’) at the same rate of false-positives and false-negatives. As always, the detection tools will lag behind. Will that stop people using the AI to generate content? No. A quick pass to insert imperfections and onto the bookshelf it goes. Will most readers care if their content is generated by AI or humans? Probably not. Compared to certain authors I could name, the AI content might be better.
Am I worried?
No. What’s the point? Despite the AI author panic, it’s going to happen anyway. The US won’t legislate against AI and China will use it for its own agenda.
It could be worse, à la Rings of Power: the elves are coming to take our jobs.
I for one welcome the arrival of our robot overlords.
You don’t even get on with your own toaster. 🙂
How long does it take an AI to write a novel? How long has it taken you (not counting the unfinished ones from 25 yrs ago?)
Who said ‘You can’t rush quality?’ Oh yeah, you did. 🙂
Genre fiction is very formulaic. How long before an AI knock-off of Dan Brown is indistinguishable from the real thing?
How do you know Dan Brwon isn’t a deep fake already? Have you seen him interviewed?
If AI’s can beat us at chass and Go already, can’t be too long before they write better novels than us.
Genre fiction is entirely within a rules-based formula. Scrape enough ebooks to build a big enough language model, you can distill fiction down to kit-form. You won’t even need an editor in a couple of years.
Is no one going to leap to the defense of human artistic endeavour?
Have you read ACOTAR? Or anything by Sarah J Maas?