Does the wave of AIs with natural language capabilities herald the end of fiction as we know it? If you think that the question is absurd, you haven’t really got to grips with the sorts of things that GPT and other generative AI software can do.
The American science fiction magazine Clarkesworld has the following message on their website:
Statement on the Use of "AI” writing tools such as ChatGPT
We will not consider any submissions written, developed, or assisted by these tools. Attempting to submit these works may result in being banned from submitting works in the future.
Editor Neil Clark announced in February they were temporarily closed to submissions because of the sudden 38% increase in what they called ‘AI spam’ submissions. He says he can spot them because there are some obvious patterns. Sheila Williams said Asimov’s experienced a similar increase. She says ‘The people doing this by and large don’t have any real concept of how to tell a story, and neither do any kind of AI. You don’t have to finish the first sentence to know it’s not going to be a readable story.’
Here are some questions to think about regarding the future of fiction, with a stab at the answers based on my current understanding:
Who owns the copyright of AI-generated fiction.?
Well, either no one or the humans who wrote the material that the AI was trained on. The words aren’t actually copied from anywhere. The AI generates words which it has statistically predicted from the Web as a response to an instruction and rephrases the material. The words have never been put together in exactly that combination before.
Who can pretend they have actually written a work of fiction that was generated by an AI?
Er… anyone can. Maybe at this still relatively early stage of these AIs, you can tell (or think you can tell) that the work wasn’t written by a human, but you could never be absolutely certain. (And the AIs will only get better.) Ironically, one criteria you could use is to see if the work has no grammatical errors or incorrect word usage—but of course, that could also be a sign of a highly competent writer. Plagiarism-checking software has been around for some time, so some sort of AI-identifying software is possible, but will it be good enough to keep pace with improvement in AIs?
Can AI-generated fiction be truly original?
Science fiction writer Ted Chiang sees a human author's first draft as often an original idea, badly expressed, while the best an AI can generate is a competently expressed, unoriginal idea. One way to look at what the AI does is create an amalgam of photocopies of original documents. Originality is always lost through the photocopying process. On the other hand, there are also lots of unoriginal human fiction authors around. Being truly original is a high bar that few writers achieve. And yet, each human’s life is unique, so the perspective they bring to a work of fiction has originality as its seed. The best writing comes from skilled writers who can turn that unique perspective into fiction. An AI can’t do that.
Can AI generate fiction that ignites an emotional response in the reader?
Authors create emotional responses through combinations of words. If you ask an AI to write a horror story, I suspect it can glean how to achieve the fear effect with all the sources available to it. Can it create fiction that makes the reader laugh? Maybe: puns can be generated, jokes have structure, certain situations are inherently funny. What about making them cry? Who knows? I’m sure someone will put that to the test. Can an AI generate the sense of awe that a great work of science fiction or fantasy can? I suspect not, but then as a writer of speculative fiction I’m biased. The most powerful emotional response to a work of fiction, however, goes deeper than fear, humour. sadness or awe, it’s the sense of human connection that the reader feels with the words spun by the writer. An AI can’t do that.
How much instruction would a writer need to give an AI for it to generate a publishable work of fiction?
Minimal instructions will most likely result in a bland, uninteresting work that wouldn’t make it to publication. However, if you are a meticulous planner, then including detailed plot outlines, world build and character descriptions could result in something publishable. The 'garbage in garbage out' rule applies. The success or otherwise depends on the quality of the input. As every planner knows, there is an enormous amount of time and effort needed before the writing begins, so the success or otherwise will depend on all that work. If you are more of a panster (that is, fly by the seat of you pants), then the joy of the writing is in the discovery, so you probably would see any point in using AI to write fiction. Either way, the AI may end up being some assistance to some fiction writers, but it can’t replace them.
What are fiction writers feeling about the potential impact of AIs?
They’re not happy. I’ll give just one quote. This is from British-Australian dark fantasy author, Alan Baxter: ‘In a world where people are still cleaning toilets and working in mines, I can’t believe we’ve got the robots making our art and stories. I thought robots were supposed to do the shitty jobs to allow more people to pursue their passions. AI is simply a hideous and well-focussed encapsulation of capitalism at the expense of humanity.’
What’s the most optimistic outcome of AIs for fiction writers?
AIs could end up being support for fiction writers rather than making their life harder. At some point it will be possible to feed an AI your novel manuscript and ask it to generate a synopsis of your novel to a set word count, taking on a burden that many writers loathe. If it could, for example, provide a narrative description f a particular place at a particular time in history, it would short-circuit a considerable amount googling and other research. The AI could also provide some required technical information in natural language specifically catering to non-specialist readers which can be adapted for a work.
We all know the definition of science fiction is notoriously hard to tie down. One way of looking at it is as the form of fiction that speculates on the impact of actual or imaginary science and technology. While the world heralds the latest technology and grapples with its consequences, the science fiction world has already explored its potential impact on individuals and society.
With the recent explosion AI systems into our awareness, we are in a truly science-fictional moment in history: here is technology that science fiction writers have speculated about for decades. While this is either ecstatically or shockingly new for most of the world, to science fiction readers and writers it’s a matter of reality catching up to our imaginations. As a collective, we’ve already thought about where AI could take us.
With anything startingly new, there are the true believers, nay-sayers and head-in-sanders. For those of us into science fiction, this is simultaneously an ‘Oh, wow!’ and ‘Oh, shit!’ moment. We can see both sides. We’ve already pictured the potential future in our imaginations. So, let’s prepare ourselves for the exploration.
We live in interesting times.
This blog appeared in a different form as the Editorials for Aurealis #158 and Aurealis #159.
The American science fiction magazine Clarkesworld has the following message on their website:
Statement on the Use of "AI” writing tools such as ChatGPT
We will not consider any submissions written, developed, or assisted by these tools. Attempting to submit these works may result in being banned from submitting works in the future.
Editor Neil Clark announced in February they were temporarily closed to submissions because of the sudden 38% increase in what they called ‘AI spam’ submissions. He says he can spot them because there are some obvious patterns. Sheila Williams said Asimov’s experienced a similar increase. She says ‘The people doing this by and large don’t have any real concept of how to tell a story, and neither do any kind of AI. You don’t have to finish the first sentence to know it’s not going to be a readable story.’
Here are some questions to think about regarding the future of fiction, with a stab at the answers based on my current understanding:
Who owns the copyright of AI-generated fiction.?
Well, either no one or the humans who wrote the material that the AI was trained on. The words aren’t actually copied from anywhere. The AI generates words which it has statistically predicted from the Web as a response to an instruction and rephrases the material. The words have never been put together in exactly that combination before.
Who can pretend they have actually written a work of fiction that was generated by an AI?
Er… anyone can. Maybe at this still relatively early stage of these AIs, you can tell (or think you can tell) that the work wasn’t written by a human, but you could never be absolutely certain. (And the AIs will only get better.) Ironically, one criteria you could use is to see if the work has no grammatical errors or incorrect word usage—but of course, that could also be a sign of a highly competent writer. Plagiarism-checking software has been around for some time, so some sort of AI-identifying software is possible, but will it be good enough to keep pace with improvement in AIs?
Can AI-generated fiction be truly original?
Science fiction writer Ted Chiang sees a human author's first draft as often an original idea, badly expressed, while the best an AI can generate is a competently expressed, unoriginal idea. One way to look at what the AI does is create an amalgam of photocopies of original documents. Originality is always lost through the photocopying process. On the other hand, there are also lots of unoriginal human fiction authors around. Being truly original is a high bar that few writers achieve. And yet, each human’s life is unique, so the perspective they bring to a work of fiction has originality as its seed. The best writing comes from skilled writers who can turn that unique perspective into fiction. An AI can’t do that.
Can AI generate fiction that ignites an emotional response in the reader?
Authors create emotional responses through combinations of words. If you ask an AI to write a horror story, I suspect it can glean how to achieve the fear effect with all the sources available to it. Can it create fiction that makes the reader laugh? Maybe: puns can be generated, jokes have structure, certain situations are inherently funny. What about making them cry? Who knows? I’m sure someone will put that to the test. Can an AI generate the sense of awe that a great work of science fiction or fantasy can? I suspect not, but then as a writer of speculative fiction I’m biased. The most powerful emotional response to a work of fiction, however, goes deeper than fear, humour. sadness or awe, it’s the sense of human connection that the reader feels with the words spun by the writer. An AI can’t do that.
How much instruction would a writer need to give an AI for it to generate a publishable work of fiction?
Minimal instructions will most likely result in a bland, uninteresting work that wouldn’t make it to publication. However, if you are a meticulous planner, then including detailed plot outlines, world build and character descriptions could result in something publishable. The 'garbage in garbage out' rule applies. The success or otherwise depends on the quality of the input. As every planner knows, there is an enormous amount of time and effort needed before the writing begins, so the success or otherwise will depend on all that work. If you are more of a panster (that is, fly by the seat of you pants), then the joy of the writing is in the discovery, so you probably would see any point in using AI to write fiction. Either way, the AI may end up being some assistance to some fiction writers, but it can’t replace them.
What are fiction writers feeling about the potential impact of AIs?
They’re not happy. I’ll give just one quote. This is from British-Australian dark fantasy author, Alan Baxter: ‘In a world where people are still cleaning toilets and working in mines, I can’t believe we’ve got the robots making our art and stories. I thought robots were supposed to do the shitty jobs to allow more people to pursue their passions. AI is simply a hideous and well-focussed encapsulation of capitalism at the expense of humanity.’
What’s the most optimistic outcome of AIs for fiction writers?
AIs could end up being support for fiction writers rather than making their life harder. At some point it will be possible to feed an AI your novel manuscript and ask it to generate a synopsis of your novel to a set word count, taking on a burden that many writers loathe. If it could, for example, provide a narrative description f a particular place at a particular time in history, it would short-circuit a considerable amount googling and other research. The AI could also provide some required technical information in natural language specifically catering to non-specialist readers which can be adapted for a work.
We all know the definition of science fiction is notoriously hard to tie down. One way of looking at it is as the form of fiction that speculates on the impact of actual or imaginary science and technology. While the world heralds the latest technology and grapples with its consequences, the science fiction world has already explored its potential impact on individuals and society.
With the recent explosion AI systems into our awareness, we are in a truly science-fictional moment in history: here is technology that science fiction writers have speculated about for decades. While this is either ecstatically or shockingly new for most of the world, to science fiction readers and writers it’s a matter of reality catching up to our imaginations. As a collective, we’ve already thought about where AI could take us.
With anything startingly new, there are the true believers, nay-sayers and head-in-sanders. For those of us into science fiction, this is simultaneously an ‘Oh, wow!’ and ‘Oh, shit!’ moment. We can see both sides. We’ve already pictured the potential future in our imaginations. So, let’s prepare ourselves for the exploration.
We live in interesting times.
This blog appeared in a different form as the Editorials for Aurealis #158 and Aurealis #159.
RSS Feed