Are you concerned about what AI–artificial and augmented intelligence–might mean for writers and publishers? Or excited by its potential? Either way, if you're an independent author or self-publishing service, ALLi, the Alliance of Independent Authors, wants to hear your ideas, insights, thoughts and recommendations. This is our call for comments on the topic of AI for Indie Authors.
Does thinking about AI make you feel like you're banging your head off an exploding wall? It's all moving so fast, it can be hard to keep up, and even harder to understand what any of it might mean for you as an author and publisher.
Psychology tells us that when things are changing fast, anxiety levels rise. That's what's happening now in parts of the author and creative community with regard to AI. It reminds me of the early days of e-books or self-publishing. Everyone thinks it's the best or the worst thing ever (remember when publishing execs used to warn “we're all going to drown in a tsunami of CRAP!!!”?), but it doesn't really matter what you think. It's happening, with or without you.
Whether you're an individual who's afraid the machines are coming to take your livelihood or an organization concerned about what it means for labor or copyright laws, you have little or no control here. While academics and governments are talking, the creative and tech industries are getting on with it, in every sector.
Those who are excited by AI and what they can create with it, and around it, are working on a myriad of projects. And publishing is no exception.
As independent authors—writers who self-publish books and selectively license our intellectual property–we need to be informed, not opinionated. Just like with e-books we need to figure out ways to work within new tech, rather than trying to legislate around it, or set ourselves in opposition to it.
This is already happening. If you want to know more, follow our Enterprise and AI Advisor, Joanna Penn, who is keeping authors up to speed in the AI section of her website and podcast and whose book Artificial Intelligence, Blockchain, and Virtual Worlds is a terrific introduction to the topic.
ALLi now wants to develop an informed assessment of the potential impact of AI on how independent authors write, publish, and trade now and into the future. We seek to do this through an open and assessed conversation with members, advisors and other interested parties. From this assessment, we will develop practical and ethical guidelines for our author and partner members.
ALLi is based in the UK and has been following with interest as the World Intellectual Property Organization held its first Conversation on IP and AI, bringing together member states and other stakeholders to discuss the impact of Al on IP policy, globally. We made a submission to this body last Feb 2020 . You can read the government response to these submissions here.
AI and Indie Authors: Call for Comments
This call for comments is the first step in that process.
Below we provide some headings, topics and questions for consideration but please feel free to respond with any relevant ideas, information, insights, comments, feedback, or passing thoughts. Nothing is too small or large.
Ideally, leave your response as a comment below, which will allow us to have dialog and debate right here on this page.
But if you'd prefer to make a private submission, you can also send your comments to Sarah at [email protected] using the subject-line “AI and Indie Authors”, saying that you'd prefer to be anonymous.
Please include as much evidence as possible to support your comments and responses.
AI and Indie Authors: A Brief Introduction
Today, artificial and augmented intelligence and machine learning systems (what we refer to in this post as “AI” or “AI systems”) are creative and communicative, rational and analytical, independent and autonomous, efficient and accurate, and capable of assigning free choice from alternatives.
This has implications for all authors, as the use of AI systems in the production of creative works grows ever more common in what is being called “the 3A era” of advanced technology and the “Self-publishing 4.0” era of independent author-publishing.
More and more authors, publishers and other creators now use, and want to expand their use, of AI tools to generate text, as working drafts, and completed works.
Yet there are copyright issues to be considered, most notably ownership and accountability.AI systems can autonomously generate creative works which would, were they created by humans, be eligible for copyright protection. Who “owns” and gets paid for AI generated books? The owners of the Natural Language Generation systems (whose AI must be trained on a dataset created from writers' works)? Or those writers who provide the text?
And how is payment to be apportioned?
To an AI, words are “data” and they are data-hungry! Their performance relies heavily on the size of training data available and data related issues are the reason most great AI ideas fail to happen i.e. the data collection process is too difficult and time-consuming.
If it's hard to access words for datasets legally, it's likely that some AI systems would use copyrighted work without permission, as evidenced by the prevalence of pirated media in markets where it is not available to consume legally.
Since NLG systems like GPT-3 are creating “original” works, it will be impossible to know what has been infringed. Therefore, the easy availability of a licensing model for “in copyright works” might encourage legal usage?
There are three routes that copyright legislators can take when considering creative works generated by an AI system:
- deny copyright protection for works that have been generated by an AI system
- attribute copyright of such works to the creator of the AI system
- attribute copyright of such works to the creator of the AI system and the (already copyrighted) creators of the works used to feed the system
We favour option 3, while acknowledging the challenges involved.
We conclude this short introduction to this huge topic with three concerns. There are many more and we look forward to hearing from you on those–and of course to any solutions you can offer.
- No doubt, there are already books for sale on Amazon and other publishing platforms that have been generated by AI. These include books that are labeled as such, for example, ALLi partner member Booxby.ai lists all their books as AI-assisted. But many more have a real or false author name on, without declaring how they were actually generated.
- Translations are another concern, as a translation of a book should also reward the original creator. AI translation tools, such as Deepl.com, can create a decent translated version of a full-length book within seconds. That can easily be human edited and published. and given that digital publishing is now international, the owner of the original work may never even know about it, let alone benefit from it.
- Because of copyright issues, natural language processing (NLP) and generation (NLG) models like Open AI’s GPT-3 are being trained only on out of copyright works, mostly books published 50+ years ago–therefore mostly older white, male, English-speaking authors leading to bias and lack of diversity within datasets, generating in racism, sexism, and other exclusionary language by the NLG systems.
AI and Indie Authors: Some Questions
- Should content generated by AI be eligible for protection by copyright or related rights?
- If so, who should benefit from this protection, and how long should it last?
- How do you feel about AI platforms using your books to create AI systems?
- No financial model currently exists for authors and publishers to license copyright works to AI platforms. Would you be in favor of this? How might such a model work? What sort of license might enable an author to benefit from the Natural Language Generation models used to generate new works?
- How will readers know if a book is created by AI or not? Does it matter?
- Is there a need for greater clarity about who is liable when an AI infringes copyright?
- Does copyright already provide adequate protection for software which implements AI? Is new licensing law is necessary to take account of the growing use of AI in publishing or are current copyright legislation sufficient? Is there a need to clarify existing exceptions, to create new ones, or to promote licensing?
- Do we need a data licensing clause in contracts?
- Can governments legislate for this? Do we want them to? What sort of protection do we want from copyright law?
- Do you currently use any AI tools in your writing?
- Can you envisage writing challenges that are solved by AI? Writing processes that are speeded up?
- How do you feel about AI in writing? Are there ethical or practical issues you'd like to raise?
- Do other issues need to be considered? What are they?
- Trade publishers are increasingly including a clause in their rights licensing contracts, saying that the contract covers “…all formats existing now and to be invented”. Do we need to resist this?
- Will “human-made” become a selling point? What can humans do better than AI? What do humans offer that AI can't?
- How do you feel about AI in publishing? Are there ethical or practical issues you'd like to raise?
- Do other issues need to be considered? What are they?
We look forward to hearing your thoughts!
IARPA awarded research grants to six firms to aid in the completion of HIATUS since it is seen as a significant issue.
There are two ways that copyright law can treat works where human interaction is minimal or non-existent. It may deny copyright protection to computer-generated works, or it may assign the authorship of such works to the person who created the program.
should grant new works generated by intelligent algorithms with little or no human intervention
I love these questions, Orna!
You raise a number of important issues. AI, once it’s fully realized as a creative tool for writers, is poised to bring a sea change in the craft — not unlike digital editing tools and digital cameras brought to photography or similar advances brought to creatives in music, architecture, and so many other fields. We get to decide how we react to it, embrace it, and grow with it.
As a creator of an AI creative writing tool (Sudowrite) I’m biased to see the opportunities. But I was a writer before Sudowrite came to be, and the project grew from my interest in understanding and learning to use AI tools in my own writing.
Writing is a challenge for many of us. Some of us love to edit and revise, but are loath to put that first word on paper. Others are the complete opposite. Some get mired in the middle, others tear their hair out landing the ending. My belief is that just as spelling and grammar checking software lifted some of the load off our shoulders, software like Sudowrite can do the same when we’re stuck in a scene, struggling to describe an object or a setting, or just not sure how to inject something fresh into a tired plot line. In the end, I’m still the writer. I bring my experiences, my history, and my taste to the subject matter. And I need to make the choices and do the work that imbues my piece with pathos and meaning.
There’s a lot to figure out here, especially with regard to rights issues. In my opinion, the most important thing is that writers’ work be protected. And as you point out, we need to continue to look for opportunities everywhere.
In a conversation with Joanna, she suggested the idea that multiple authors could license their related works (say, thrillers) to train a model that could then be licensed to others to help them write thrillers of their own. I love this kind of creative thinking, and I think it’s a great example of how we can best react to changes that come with any new technology. It’s not just about how to protect ourselves, but also finding innovative ways to grow and benefit from the changes afoot.
Orna and Joanna, this is such a great breakdown of the fast-moving changes in the space. Too often, authors come to the subject with the mindset that AI is some scary outside force that’s coming in to displace them. We all know the AI wave is coming — indeed, it’s already here with search, Amazon, Facebook and all the rest — so it’s important to get out in front and make sure we make use of AI to help advance our careers and the self-publishing movement.
When Dr. Matthew Jockers (co-author of “The Bestseller Code”) and a group of authors launched Authors.ai (which recently became an ALLi partner member) some 18 months ago, we made sure to send the signal that everything we built was in the tradition of Grammarly, Pro Writing Aid and other tools that authors and writers use to improve their storytelling. A fair number of authors understandably are skeptical of the notion that AI can help them with their craft, and that’s fine. If it envelops our sector with no accountability, the potential for harm is considerable, and Joanna has masterfully laid out some of the scary scenarios in her podcast (and no doubt in her book, which I just bought!). But the answer is to help steer AI down the right path, not to shut our eyes and hope it goes away. Because if we do that, then the worst use cases will come to pass.
What we’ve found over time is that thousands of authors are now using our artificial intelligence to good effect, as evidenced by the testimonials and emails we receive all the time, not just from authors but from book editors. Others don’t want anything to do with AI, and that’s fine. There are many ways of writing a great novel, and AI has just come onto the scene. So let’s learn from each other and put down those markers to clearly define what crosses the line when it comes to AI and what doesn’t cross any ethical line but instead helps authors with their writing journey.
The copyright issues are complex as authors typically have minimal input and few realistic recourse options available due to the high cost of litigation and fly by night nature of the infringers.
But your suggested approach is sound. The biases on multiple grounds in older works in the English language is a definite issue, which needs to be watched for carefully.
Authors are now competing with algorithms. We need to step up. Becoming creative across multiple disciplines is one option. Our experience with creating NFTs for authors shows that there is a capacity among authors for creating multi media support material for books.
The involvement of indie authors in cover design/selection, book trailers, and multiple other creative outlets is a part of this process of authors also evolving.
The AI engines will have to be good to keep up. And can we please get Amazon to tag AI produced books so we know what we’re up against?
Thanks for opening up the debate, Orna. It’s so important that creatives engage with AI as it is not the future — it really is right now!
Things are changing fast in the AI space and with Microsoft licensing Open AI’s GPT-3, AI tools will rapidly become part of every writer’s process regardless of whether they are aware of it. A Chinese language model 10x bigger than GPT-3 was launched this week and things will only become more complex over time.
Of course, we also use AI tools right now — Google search, Facebook, Amazon algorithms, and more.
In terms of AI in creativity, musicians and visual artists are embracing new tools and in writing, we have an expanding array of options.
I look forward to hearing people’s thoughts.
An excellent overview befitting ALLi’s members. For the many ALLi fiction writers, perhaps consider reading a more philosophical, humorous implication of AI in the best-selling book and 2019 Booker Prize longlist, Frankissstein by Jeanette Winterson. It won’t help writers face the pragmatic challenges of machine v human writing, but it will satisfy the funny bone.