ALLi recently ran a short survey asking members four questions about AI: how you use it, whether you have a policy, what worries you most, and what help we could give. Here are the results and our next steps.
We expected a few hundred replies to our survey on AI but we received more than 1,200, along with a total of 430 free-text comments, several detailed enough to be blog posts in themselves, and one very interesting manifesto.
A few members felt the survey itself was biased. We had asked respondents to categorize their use into minimal, assisted or integrated, which is how we categorize our policy responses, and some felt the category of “minimal” was not minimal enough. Others felt the opposite, and that the tone of the survey leaned anti-AI.
As a collective voice for self-publishing authors it’s impossible for ALLi to please everyone on this: opinions are too diverse and, in many cases, hotly held.
Our aim is always to protect authors’ rights and freedoms and with AI, as with everything, we aim to be inclusive and transparent.
Drawing on our mission of ethics and excellence in self-publishing, we encourage our members to each develop their own AI policies, and equally to respect the choices of others. You can find out more about our own policies here: AI policy
Thinking about your own AI policy?
Our free guide, Writing Your AI Policy, sets out where ALLi stands on AI and includes a sample policy you can use as a starting point, whatever your approach.
The Survey Results
What follows is a summary of what you told us. Think of it as a snapshot. Things are changing fast, especially in this arena.
How Members Are Using AI
Almost half of respondents (48 percent) described themselves as AI-minimal: they use only the most inescapable, basic AI tools and prefer physical processes and assistance that is human.
Nearly a quarter (24 percent) used AI assistance for select tasks in both publishing and writing, developing ideas, editing and research but not drafting.
A further 16 percent confined their AI use to publishing activities only: market research, design, database work.
Just 5 percent described themselves as AI-integrated, using it across every process.
And 6 percent were still figuring it out.
Written AI Policies: Most Members Do Not Have Them
Only one member in 10 (10 percent) currently had a formal AI policy on their website. A third of members (33 percent) intended to write one. The majority, 58 percent, had none, and weren’t sure why they should.
Interestingly, members who stated their use was minimal were the most likely to have a policy (13 percent), and the most likely to be planning one (39 percent). Those who used AI most were the least likely to have written a statement about it.
From your feedback, it was clear that in many cases, the reason is caution rather than concealment. In a climate where disclosure can invite hostility, many authors who use AI for marketing or research have decided that saying nothing is the safest approach. Being afraid to speak up is not a healthy state of affairs for anyone, especially authors, and it indicates where ALLi can perhaps be useful, which we come to below.
What Worries Authors Most
Asked for a single biggest concern, members spread themselves widely. The largest slice (22 percent) chose concerns beyond publishing altogether: the environmental and economic costs of the technology. Close behind came protecting one’s work from being used to train AI (21 percent). Then the impact on the wider author community (10 percent), reader reaction (8 percent), copyright and legal issues (8 percent), and being left behind (7 percent).
Publisher or agent reaction barely registered (1 percent), which for a survey of happy independent authors isn’t surprising.
One interesting trend was the concerns correlated with usage.
AI-minimal authors worried about wider harms in the world (27 percent) and theft for training (also 27 percent).
AI-assisted authors worried more about people. They observed that readers and other authors react sharply once you admit to using the tools.
And among the small AI-integrated group, the single largest concern (21 percent) was the reaction of other authors.
Not sure what an AI policy should say?
A policy is simply an honest statement of how you work. Our free guide includes a sample policy to help you describe your own practice in your own words.
Reading Between the Tick-Boxes
Nearly a fifth of respondents (18 percent) chose “Other” for their biggest concern, and the 225 comments they left are the most instructive part of the survey.
There were six main currents.
The first, and strongest, was the flood: that AI-generated books are, in the words of many, “clogging”, “swamping” or “drowning” the retailers, and human-made work is losing discoverability to volume uploads that can be produced in an afternoon. Travel writers, non-fiction authors and genre novelists all reported this.
The second is the erosion of human creativity and thought, from “brain rot” to “outsourcing our critical thinking”.
The third is ethics: the training question, and the conviction, held with real force by some, that any use of these tools is complicity with bad actors.
The fourth current ran the other way. A substantial number described harassment, accusation and “witch-hunts” from fellow authors, sometimes triggered by a single em-dash. Several commenters said that no AI-assisted author they know now feels safe to disclose use.
The fifth is fraud: the torrent of AI-driven scams impersonating publicists, publishers and even known authors, arriving within hours of a book’s release.
The sixth is simple accuracy: hallucinated facts and confident wrong answers.
Many respondents remarked here that the question should have allowed more than one category because they wanted to tick them all. That’s a fair comment. All the same, we also wanted to ask you to prioritize, as we felt that would give clearer data.
How ALLi Can Help
Asked to choose one action ALLi could take, a quarter of members (25 percent) chose an advocacy campaign promoting author rights. Next came guidance on protecting work legally (18 percent), then a sample AI policy that can be adapted (15 percent), clearer information on platform disclosure rules (8 percent), an AI licensing framework (7 percent), community discussion (6 percent) and understanding the copyright position in different territories (4 percent).
Another 17 percent chose “other”. Once again, their comments show the nuances of the issue.
The pattern by usage is striking.
AI-minimal authors showed a marked preference for advocacy (40 percent).
Authors who use AI for both writing and publishing asked for sample policy language (25 percent).
The group who described themselves as AI-integrated asked most frequently for community discussion (17 percent) and, in their comments, 33 percent of them independently, over and over, said: help us disclose without being punished for it.
These requests fell into three clusters. The majority asked for education: practical, small-bite guidance on what these tools can and cannot do, especially for business and marketing, and “unbiased” information rather than fear or hype.
Second most popular was a call for labeling, regulation and clearer definitions, so that a reader can tell a wholly AI-generated book from an author-directed one and so that authors know what counts as “generative”.
And third, we had a request that went in two directions at once, a call for ALLi to take a stance. Some respondents wanted an unambiguously anti-AI organization; others wanted us to defend members who use the tools from what they describe as a mob.
What We Take From This
The survey gave us lots to think about. For now we are focusing on three big takeaways.
Advocacy for author rights remains the common ground. Whatever your position on use, almost everyone wants their work protected from unlicensed training, wants the flood of low-quality generated books labeled or restrained, and wants a say in how the licensing question is settled. This is where ALLi has put most of its work to date and will continue to put its weight. We are strongest in the areas where the community can speak with one voice.
The community itself needs tending. Reaction from other authors is a real fear, and the main one among members who use AI most, and the harassment described in these comments does not make anyone safer or any book better. ALLi’s Code of Standards asks members to treat one another with respect and to disclose honestly. As well as the work outlined above, we’ll be looking at how we can support respectful disclosure and how we respond when members are targeted. We may need your help to do this well.
ALLi always prioritizes giving good information. Some 10 percent of you have a policy; a third want to write one; most of the rest don’t yet see the need for it. At the same time, a sample AI policy was among the most-requested items, and the top request from authors who use AI in both writing and publishing. (You can download it using the link in this post.) Also requested was clearer guidance about individual platforms and plain explanations of the legal position in each territory. We will be bringing you more of that.
Thank You, and Keep Talking
Obviously a survey like this can’t settle the many questions AI raises, and it wasn’t meant to. What it can do is show us to ourselves. Thank you to everyone who took the time to answer, and especially to those who wrote at length. We read every comment, including the one that called the whole exercise stupid!
The conversation continues on the member forum. Do keep telling us what you think.
Writing Your AI Policy: free guide
Set out how you work, in your own words
Whatever your position on AI, a clear policy tells readers, collaborators and retailers where you stand. Our free guide explains where ALLi stands and includes a sample policy to use as a starting point, in your own voice.
Not an ALLi member yet?
ALLi is the global non-profit association for indie authors. Members get advocacy on author rights, all our guidebooks free, personalized advice, and a community of thousands of working indie authors.