AI - THOUGHTS AND SPECULATIONS REGARDING BOOKS AND COPYRIGHT

I'm finding that few days pass now without me being compelled to read an article about the consequences and hazards of AI. But, speculating exclusively in terms of my profession (writer/publisher) and of books and copyright, the magnitude of future consequences for writers I am only just beginning to comprehend. The reaction thus far, from my profession, appears to be comprised of equal parts mystification, apathy, wishful thinking, patience, reticence, resignation, disbelief and fury.

I'm also struggling to identify many benefits for writers, short or long term. Advances in a foolproof automation in the production and distribution of books, will surely be massively outweighed by the widespread infringement of copyright that is already descending. My fear, looking forward, is of a proliferation of near infinite transgressions against copyright/intellectual property, if the theoretical capabilities of AI are realised. And this is merely the use of the gimmicky end of the tech that anyone can use, for free.

Whether books are anticipating mimicry and exploitation for commercial gain, or novelty and personal amusement, I wonder if the differences between books generated by humans, and those by software, will become indistinguishable in the near future.

My wife and I asked one of the fledgling AI apps, made accessible like free candy, to write a story in the style of Adam Nevill and it quickly produced something that a ten year old child might write. Don't attempt this. I conducted an experiment because it's my work that is at potential risk of unauthorised exploitation, and I needed to assess the degree of counterfeiting that was currently possible (beyond eBook piracy that has plagued me for years), with the available software in the public sphere. I don't want to see (anymore) AI crap generated, that is related to my stories, so I'm not setting any precedent here. DON'T.

What was produced was an outline of a story, not a developed work of fiction. A treatment for a plot that was instantly familiar. An attempt, in fact, by the software to rewrite my short story 'Where Angels Come In'. Characters, location, situation and plot were the same. A bland and infantile rendition, maybe, though one strangely hyperactive in tone that was eerily human, in a geeky, childish, tech voice. I still went cold. I had a sense that this was the cave painting that will precede the capability for a far greater sophistication, that will evolve in a few years, not centuries, or even decades.

This short story is available, along with eight others in free eBooks, that have been, I suspect, scraped and recycled. This suspicion was further strengthened by the thousands of downloads of the free books, that suddenly occurred within 24 hour periods, in June and July, from Amazon. The eBook has been available for free on Amazon for seven years, but thousands of random downloads occurred within days this summer. What other variable, beyond AI during its current scraping frenzy, could account for this? I've read that similar scrapings have happened at Smashwords and at a torrent site offering a few hundred thousand illegally distributed works of fiction. It's possible that digital editions of all of my published books, have already been scraped and are "in the machine" as it were; all ripe for some form of future access and usage and amalgamation into other entities, without my permission. And my permission will never be given.

The AI foetus in the app that we used is merely dreaming at this stage; when it's a full grown adult and using the equivalent of a word processor, what will the implications be for the very idea of authorship and copyright? I don't feel that I'm seeing enough of a protest, let alone an investigation, from the publishing industry.

I'm also made to think of what creates a book. I'll mention a little of what becomes a book. A growing mastery of language and sophistication in expression, gained after decades of careful reading and study. A combination of personality traits, experience, observation, deep thought, to produce a unique perspective and processing of one's own time and thoughts. A deep mining of the sub conscious mind. Years of practice, research, the exhaustion of rewriting the individual sentence and the endless wrestle with recalcitrant syntax. The cultivation of an aesthetic and lexicon, of rhythm, even meter. The desire to create, and to make sense of anything and everything, in stories. Now, no discussion about the "fair usage" of copyrighted work can be countenanced without a fundamental understanding of what is involved when a human being writes meaningfully. And yet, this consideration doesn't seem to be on the table.

The temptation that is being offered, is a temptation to embrace a lazy abnegation of an active concentration; not least the conscientiousness, dedication, investment of time and the highest powers of the mind required to write well, and to produce written work of value. We'd be removing our unique ability to communicate and commune with each other, abstractly and laterally, within that special bond of writer and reader, story and imagination. Human ability and talent seems ready to be devalued, or to merely provide fuel for the runaway tech train.

Why would we introduce the entity of AI into the field of literature? I can't think of a good reason.

Who will benefit? Publishers that don't want to ever pay a writer again? Cyber criminals? People who can't write, but fancy being authors and can have a go by typing keywords into a field, so that the software can produce "their" book, which is comprised of the work of others that preceded it? I can't envision a more passive human activity that results in creation. But this process would involve the creation of an artefact, or file, that does not belong to the keyword typist, or even the AI. It's a stateless combination of other states. We'd be divorcing ourselves, or disconnecting, from one of the most important abilities we've evolved as a species. Remember what happened when King Lear abdicated and gave his crown away.

At present, I'd guess AI's early, accessible incarnations are the equivalent of the Casio calculators and digital watches of the 70s. But even at such primitive levels, they're cause for concern. When the tech is as advanced as a modern Apple Smartphone, or PC, will any creator, whose work possesses artistic and commercial value, be spared scraping and regurgitating? AI's very first misuse of intellectual property seems to have begun with books, art, photography and music. From the inception of training the technology, writers and books were among the first victims.

And I wonder if AI, in the near future, will offer users the ability to produce "original" stories, derived from any published author that has ever existed; these would be "new" stories assembled from the many components (style, character, story structure, thematic preoccupations, vocabulary, syntax, the aesthetic within description, the very imagination, thoughts, ideas and inspiration) of any published author with a digitized body of work. Multiples of different authors and their books might also be used, in variations and combinations, to generate original reconfigurations of hundreds of thousands of existing works. Maybe in seconds.

That is not authorship, though, or writing. Anything original that arises from this form of AI grinding is, surely, authored by the software and not the chimp that typed the keywords into the field to produce the amalgamations? But the AI itself would be a pirate and doesn't have permission to use any of the material that it was trained on.

If regulation is going to remain voluntary at tech companies, then there is no regulation. If the software that enables "new" books to be assessed, to see if AI was used in their creation, is going to be abandoned, then there is no protection, or deterrent to counterfeiting the canon of world literature. That to me is the single most stupid thing that I've read about AI: no meaningful regulation, nor means of detecting its misuse (surely the development of detection should match the development of the AI tech, step for step?). Tech companies are enabling the means for anybody to commit perfect crimes, into infinity. Are the creators of the technology above the law then, and government? It's like a hideous joke, an absurd and tragic third act for mankind: to let code-writers be God.

History suggests the answers to the questions that I ask myself about AI usage, as regards copyright. I think that in every way that the internet has been used and misused, AI usage will follow in its footsteps, immediately; from porn and cyber crime, to disinformation and copyright infringement. The same flawed species will be using the new technology, and the same flawed species will be advanced, enhanced, altered and damaged by the new technology. Business as usual, but with a significantly more powerful technology that creates far more unforeseen consequences; consequences that are presently incalculable. So, if the technology is available to the general public, it will be used and misused by the general public, but also criminal elements, opportunists, commercial interests, hostile states, dipshits (never discount what the humble, loner dipshit can do, with a combination of malevolence, resentment, a mental disorder and a keyboard).

I worked in publishing when vast backlists were being digitized into eBooks, or plans were being made to do so. The initial intention of the publisher was to scrape as much profit as possible, via this new digital format that was easy to manufacture and distribute, with miniscule costs. And also, to not get left behind - even though it's problematic, we have to join in. I remember a 5% royalty on net receipts being proposed for authors and actually drafted into the early letters of agreement, on sales of eBooks with price points slightly less than print editions; a proposed agreement preceding the contract stage. This royalty rate became 15%, but the sheer unfairness of that rate, saw it climb to 25%. It remains much higher, and fairer for authors, in the indie scene. But I still wonder if raised royalty rates ever offset the losses incurred through piracy. During the eBook gold rush, another executive wanted to buy short stories at a flat fee, then break up the anthologies and sell the stories as individual books without paying the authors royalties (I threatened to walk and it never happened). But, intent and all that.

I also remember one of the many meetings that we conducted about the mass conversion process of our backlist, in which I made an argument about the inevitable future of book piracy. My argument was that no book should be digitized until it was possible to protect copyright across the board. My "Armageddon" scenarios were mocked. As an author, as well as a publisher, my concerns and arguments differed from the majority of my colleagues. I even remember a senior colleague claiming that people who read books are not the sort of people to indulge in book piracy. As flimsy an argument as the current "fair usage" argument in the US. This suggested that literate people were a class, or subsection of humanity, not predisposed to dishonesty and petty theft. A nice idea but wishful thinking. Even at that stage, and we're going back nearly twenty years, a culture of entitlement that all digital material online should be free, was deep rooted in users of the internet, if not already normalised behaviour.

So there is a cultural element to this whole argument, and the entitlement to "free stuff" has continued and accelerated across the arts. Maybe music has been hit the hardest, but I'd guess film and TV are catching up. I've read a variety of estimates of how widespread book piracy is, and the one estimate that seemed to stick with pundits in trade publications, is that around one quarter of eBook downloads are pirated. Is it that high? There are no accurate figures. But, to my eye, even decades ago, the use and consumption of pirated material had little stigma attached to it. Very quickly, it'll be the same with AI generated "entertainment" across the arts and entertainment fields. Instead of the use of copyright-infringed material being classed as socially unacceptable, it'll cause few qualms. So once AI art is accepted, and if there is no accountability for copyright infringement, do we stop counting, along with our rights of ownership to our own works, even our minds?

EBooks, however, have become an asset to writers, publishers and readers. They've been a boon for literacy and commercial publishing. Despite the piracy, I'm all for the format if it improves literacy, book sales, access to books and actual reading. But books written by AI (prompted by humans with commands and keywords), and composed from out of the myriad of scraped published works, are something else entirely. Allowing them to proliferate jeopardises our agency to react to experience, and make sense of it, to communicate, and simply tell our stories, as a sentient species. If a technology exists that allows individuals, or organisations, to recycle, glean and cherry-pick from published work, and create unique counterfeits compiled by software, then I have no doubt that the software will be used for these purposes. As soon, more or less, as it's ready.

Another issue that that has always bothered me: if the giant tech companies can stop online eBook piracy, then why haven't they? Free stuff equals traffic? And if AI has been trained on copyright infringement - so we can assume that most of its data, pretty much, has been stolen without permission - then, surely, its existing data should be scrubbed and the industry needs to start over within regulated, legal boundaries?

The fair usage argument in the US, I fear, will win - or be dragged out until everything of value is scraped and the process is impossible to reverse. Then it'll too late and the rival software creators will all have been trained on the same cosmos of scrapings. What happens next to books will be out of everyone's hands. That, I suspect, is either an ideal situation for those creating the technology, because the captive audience will be monumental (and will soon have to pay to use the best apps); or the trashing of copyright will be considered an unfortunate by-product of progress. When AI starts devouring other AI creations and reproducing new AI creations, that will be, in turn, devoured, how would the origins of any written material be discernible?

Consequences and More Speculations:

Let's say that future AI software becomes sophisticated enough to write a new novel by Tolkien. Then, surely, there will be attempts to upload and sell the myriad results of new stories "in-the-style-of" [add any popular author's name], produced by the technology and its so called "authors" and "publishers". And immediately.

In effect, will these new "authors" be little more than those who fed real author names and book titles into fields and hit "enter"; or even fed the precise measure of ingredients - the actual epub files of specific authors - into the grinder. Is that a thing, or going to become an issue? Because then, the counterfeiting will be more specific and focused (and easier to detect).

Initially, I suspect that much time will have to be invested in reorganising and editing the first generations of AI generated results (but this is not writing; it's the equivalent of turning plant matter and chemicals into cocaine). But I wonder if, eventually, the AI generated books will need much editing at all, or any. I don't struggle to imagine the existence of AI/AGI that will be able to produce professionally composed and edited books, that will appear to have been written by the actual authors whose scraped books trained the AI. New works may also be composites of the work of hundreds of authors.

A great many people are going to see this technology as a great opportunity. Writers don't have the litigious might of Disney. They're easy pickings.

But soon after it becomes so easy to type [add myriad book titles and author names here] into a field, seemingly new and original works will be produced in their millions, annually - particularly if AI is also used for super fast file creation and distribution (believe me, if you want to publish a book properly in several formats, that process is fraught and time intensive). The proliferation of AI generated books will be impossible to regulate. Compliance will be guidance at best. It'll be obvious to spot new poems by Ovid and new novels by Joseph Conrad, but not as easy to identify precisely which books are AI generated within the torrent of new thrillers and genre titles. And they will flood.

How will a retailer offer any meaningful compliance or regulation to police AI generated books? How would they know how new titles were produced? How would they test each book - will that screening capability ever exist in effective form? What if the "book" is only published in print editions?

When AI generated books attain significant commercial value, the game changes again - I suspect that traffic and sales will decide the argument of what is permissible and what isn't.

An equivalent of take-down notices, currently used to remove pirated copies of books from torrent sites (a tedious process that I continually practice, with mixed effects, to get my titles removed from torrent sites), will be issued in vast clouds by paranoid authors convinced that their work has been scraped and recycled into these new books for sale; books that appear uncannily similar to their own. The new titles that are commercially successful will attract swarms of plagiarism cases.

There will be several high profile legal cases in which fake AI books become international bestsellers, and the "authors" (data in-putters) will get rumbled, eventually. But they won't really be punished - the books will be taken down but those responsible will keep their ill gotten gains, encouraging so many more to try their hand at the same thing. We've already seen similar with scoundrels publishing existing books within new covers and under new author names. And don't forget the agency that was rewriting existing romance novels by altering a few words, on demand from aspiring "authors". Has anyone done time for the piracy of books, or even been fined, or been made to pay reparations to the injured parties? I suspect it's seen as a crime so soft that it doesn't warrant the hammer of the law. And who gets ripped? Only writers. Boo hoo.

There won't be enough courts and intellectual property lawyers in the universe to cope with the new cases raised by aggrieved and suspicious publishers and authors. I'm probably just going to use a crowbar and hope that I was right about the chump that produced something that reminded me of The Ritual.

Every single book published after a certain date will need to be viewed with suspicion, because it might have been generated by AI, or its creation may have been assisted by AI. How will we determine what is original and what has been wholly, or partially, recycled from previously published and copyrighted works?

Millions (more) titles will be uploaded onto retail platforms each year, and perhaps so many that online platforms won't be capable of supporting the infinite number of variations being churned out by the users of the AI technology. Remember, it may eventually produce a new novel by Jane Austen, ready to go, in minutes. Its work rate will make George Simeon's resemble idling.

Even authors who should know better, will use AI to construct and edit their books, and even have AI rewrite parts of them. There will be huge controversies and heated arguments about even partial use of AI in the authoring of books (by both new and experienced authors). These disputes will rage like forest fires. I reckon if we looked hard enough now, we could find examples of AI doctored books.

There will be so many bogus arguments about fair usage and fan fiction and new art forms, developed to excuse the new torrent of "books". Those who benefit from using AI software to create books will devise any number of arguments to justify their self-interest.

Any new book that attracts commercial or critical interest, will be scraped within the first hour of publication.

It's possible that we won't be able to identify "real" new authors. Intolerable, but how will we know which new authors are real writers?

There will be users and their AI software and its vast libraries of scraped books; the interface will be the keywords that the two entities share with each other. Will that be how new literature is created - with keywords and scraped epubs? Maybe keywords will be replaced by little animations too, because then you don't even have to type any words into the fields. Emoji heart + brawny chest + dragon + 600 pages = Epic romantic fantasy, set in Middle-earth.

Authors who established their careers before a certain date, will have the best chance of surviving.

What happens to proofreaders, copy-editors, editors, cover artists, text and book designers, audio book narrators/producers?

***

I look forward to being wrong about all of this, or even most of it. But writers should debate this now and form some consensus on ethics for publishing, authors and retailers in the AI future. Most importantly, how should the tech be regulated? We need something similar to Asimov's Law. How about:

"AI may not copy and reproduce the writing of a human being, without permission, or, through inaction, allow a human being's writing to be copied or reproduced. AI must obey orders given it by human beings except where such orders would conflict with the First Law. AI must protect and respect copyright."

***

That's quite a lot to imagine. You might suspect the presence of a protectionist tone in this post, but I'm entitled to one; I have made my life about books and writing, and believe that a great book is one of the highest human achievements. Books are sacred. So I'm struggling to watch the current, casual, baffled stroll towards the precipice.

What am I going to do? Well, I am not going to do anything differently. I'll continue to write my books using the methods that I've honed over decades. I am going to research and write my books incrementally and laboriously. I will enjoy the same euphoria of creation, and endure the same despair at falling short. Eventually, wearily, I will accept that something is as good as I can get it. I will rewrite my books until my internal editor (also forged over decades) is satisfied. I intend to hire the same professionals that I currently hire to produce the various editions of my books - artists, editors, managing editorial pros, designers.

But, as a reader, I'll not be reading stories or books written by AI, or assisted by AI editorially. If authors stop rewriting their own sentences, until they are as good as they can get them, and use AI instead to rewrite their sentences, they're not real writers. At least, I will never accept them as such.

My stories won't be appearing in collections alongside AI generated stories, or within books with AI generated covers.

If it becomes impossible for me to discriminate between books produced/partially produced by AI, then I will not read any book published after a date of my choosing. I'm guessing that date will be within the current decade. I'm sticking with human writers and I'll never run short of reading material.

I may become a pitied anachronism. So be it. But my future books will be written and created in the way that books have always been written and created. By humans.

Note - as a member of the Society of Authors, their suggested disclaimer for AI's usage of my writing, stands for everything I publish: 

"The Work of Adam Nevill, aka Adam L. G. Nevill, may not be used or accessed in any manner which could help the learning/training of artificial intelligence technologies."

[Disclaimer - please forgive any errors, inaccuracies and typos; the post was authored by a human]