AI and Creative Industries: Our AI Policy and Use of Technology

AI and creative industries

Everyone is, to use a colloquial expression, getting their knickers in a twist about AI. They are either horrified and seeking to protect themselves, or excited and dashing to engage or even invest. We thought it was time to state Castle Sefton Press’ AI policy and explain how we use technology. We don’t think AI and creative industries go together.

Use of Technology

Of course we use technology; brilliant technological developments have made a small enterprise like ours viable. Apart from the internet, our websites, all the major sales platforms and social media, we use a great piece of software called Vellum for typesetting, the super Affinity suite for cover layouts, logos and image processing and good old MS Word for editing and proof reading. Some of these third-party apps and platforms may make use of robots, but we never do directly. We turn off any AI integrations we can.

Tools, not masters

We are not digital artists, and even our book covers are initially created using physical media. (You can read about the media we use in another article). That still involves technology: pencils, easels, modern acrylic paints and brushes are all technological developments. I never ask a pencil what it thinks I should draw, or to draw something for me. I would as soon ask a computer the same questions.

That’s because technology gives us tools and not masters. No one here uses it to create content. Every word in every book and every article comes from the authors’ brains, possibly adjusted according to the human editors’ suggestions, and nowhere else.

At Castle Sefton Press we are not convinced that computers do create content. We suspect that they are doing what they have always been good at, processing data. They are just doing it much faster and more powerfully than before. That means AI has no direct role to play in the creative industries.

Machine errors …

I notice when reading that mainstream publishers, who are aways so quick to suggest that independents like us are not ‘proper publishers’, often now use machines for proofreading. This is one example of several in On Nationalism, by Eric Hobsbawn, edited by Donald Sassoon, published by Little Brown in 2021:

‘…little but common economic backwardness untied what we know about the Yugoslavs…and the very concept of Yugoslavia was the product of intellectuals in Austro-Hungary…’

This arrests the reader because it makes no sense, and should surely read ‘united’. But as ‘untied’ is properly spelled and grammatically correct, the machine left in charge cannot possibly know that it is a nonsense. I can assure you that our proofreader would never miss such a mistake. If it ever got to her after extensive perusal by the author and editor.

… and human ones

This is not to say, of course, that we never get things wrong, only that you won’t find systematic errors like this. They are due to leaving a machine to do a human being’s job. If you do find a mistake in one of our books, please let us know. The beauty of print on demand is that errors can be corrected immediately in all editions.

Worries about AI and creative industries

There is a lot of concern about AI and theft of intellectual property among creatives in the UK. Understandably so given the government’s recent enthusiasm for the idea of AI bots crawling copyright material to learn from it. To explain our views on this issue, we’ll share the story of a friend who, some years before AI was talked about, wrote an excellent biography of an important, though not very famous, British man and published it through an independent publisher.

Plagiarism

The book did reasonably well but was not a bestseller. A few years after it was released, our friend the author came across a novella presenting a fictionalised version of his subject’s story by a US author. It published by one of the five huge global publishing corporations. Not only did the fiction author not reference our friend’s book, but he claimed that the novella was based entirely on his own research during a trip to the UK. Our friend could prove this was not the case as the fiction author had used conversations with and recollections from people who had died before his visit. The only record of these was only in our friend’s biography.

Our friend wrote a courteous letter to the huge global publishing corporation, suggesting that the author might like to credit his book in future. In return, he received an aggressive official letter from US lawyers threatening him with legal action if he ever mentioned the issue again. Of course, he let it go. He could barely afford legal action of any kind, and was certain to be defeated by corporate lawyers backed by vast funds.

It’s about power not technology

The point is that theft of intellectual property is an issue of power, not technology. Those power structures were sadly in place before the advent of AI. Yes, the technology makes it much easier and faster to steal stuff, but it is only the power structures that we live within that allow it to happen.

I’m fairly sure that if an individual – Ethel Smith, say – developed an AI bot in her bedroom in a terraced house here in Stoke-on-Trent and used it to trawl and borrow from the works of high profile authors across the country, the UK government would enthusiastically support her prosecution under existing copyright legislation.

The Castle Sefton Press AI Policy

We don’t believe AI is creative, or should have a role in creative industries. Nor is it the cause of theft of intellectual property. But because we can only do our best to survive in the world we live in, we have to accept that other people will be using AI.

Castle Sefton Press has developed a robust AI policy:

• avoid smart devices as far as possible.
• turn AI assistants off wherever possible.
• deny access to our creative work for AI training when possible.
• don’t worry too much if we can’t manage any of these because of power imbalances – we can only do our best.
• allow AI bots to understand our website by use of LLMs.txt. People are going to be using AI search results and answers whether we like it or not. In which case, it’s better that they receive accurate information about us.
• use AI for routine tasks when appropriate such as creating subtitles for our social media videos which can then be checked by a human. These are computer tasks for which the software has been labelled ‘AI’.

Here at Castle Sefton Press, we’d much rather talk to the pencil than ChatGPT.

Declaration of Uses of AI

• We use AI to generate subtitles on our social media videos. These are then checked by a human (usually me!) before publishing.
• Some of our product images on the website and online may have AI generated backgrounds. We are a tiny business and cannot afford expensive photo shoots for our products. Stock image or computer generated backdrops show our items in context, but our product designs themselves NEVER use AI.