Going with the AI flow

AI is being thrown around every industry as a term that holds equal amounts of promise and horror, depending on who you ask. There’s a need to bring greater clarity to this complex landscape for the publishing sector. We need to move from publishers blandly nodding at the mention of AI to feeling empowered to make best use of this emerging set of technologies which is set to change the working landscape and help to free up publishers’ time to focus more on workforce creativity and supporting enhanced author care. 

Though individual publishers are starting to consider using AI, across other workplaces some 54% of employees have "no idea" how their company uses it, and 70% of staff using AI in their day to day work are not telling their bosses about it. The flip side is that some organisations prefer to ban generative AI outright (Space Force, Samsung).

It doesn’t have to be that way. There are particular workflows that can directly benefit from the application of AI technology, some would even require little to no investment on behalf of the publisher. 

Marketing is perhaps the most obvious. Marketing and advertising was one of the early areas of business that embraced AI in their workflows. AI can generate huge volumes of content and captions. It can quickly edit media, find optimum times to post, and measure response rates and sentiment efficiently. There are a variety of tools already in place to help marketers with utilising generative AI, but getting started is about understanding the way generative AI works, where that data comes from, and how to prompt it in such a way to get the best-quality results. 

Visual platforms such as Midjourney, Dall-E 2 and others can turn non-designer prompts into outputs that can create stunning and believable images (and words are getting there). Style Tuner, in Midjourney, enables a company to train their own AI image generation style, allowing publishers to develop a unique in-house look. 

We should work to our strengths and, within publishing houses, create human-centred approaches to working with AI 

Platforms such as VirtuLook can generate backgrounds for your books and Adobe Firefly can generate content and make editing faster and easier for those marketing pieces that don’t require a graphic designer’s touch. Logo development, colour schemes, taglines, and brand voice are just a few of the ways that integrating AI can into a marketing workflow can free up time in the making of content so marketers can focus on the strategies behind them. 

In a consumer support role AI chatbots can be integrated with marketplaces, publishers can enable users to visit their websites, chat to a bot that has been trained on their database, and accurately suggest books to read, and help the user purchase their preferred copy – within the publisher’s ecosystem. Likewise, integrated chatbots can also gather valuable feedback in ways that feel conversational, with the users building up valuable in-platform audience insight from readers who might never otherwise offer up that information, such as: “what did you think of the cover of this book?” or, upon returning to the site the bot asking “did you enjoy the last book you purchased? Would you share it with a friend or loved one?” 

There are also innumerable ways that AI can be trained on data to allow access to complex systems in natural language: “can you tell me what was the best-selling book in the last week of March of 2017, and how many copies it sold in bookshops?”, or “make me a graph of the most valuable book we have sold rights to in non-English speaking countries and pull a list of themes or key words”. These are just a couple of the potential uses, all of which come with the caveat of data security, consideration of copyright, and ensuring the AI platforms/chatbots/etc. are not adding your input data into their training databases at large. 

AI also doesn’t have the best EDI record, and what goes in to the training system is what comes out. Indeed, the Charter AI Worker Inclusion report found that ‘women, workers of color, and those over 55’ may be the most disproportionately disadvantaged by AI use in organisations. That being said, there are more AI-powered companies starting up to help organisations avoid biases and to write, expand, and focus their EDI policies. In fact, research at the University of Salford found that asking generative AI to develop and write EDI policies is promising, especially if there are existing policies from which to start.  

AI tools have the ability to help level the playing field between low and higher skilled workers, and several studies have found evidence of this. AI can also be used to make works (and working) more accessible, breaking down language barriers, powering voice technology that sounds warm and human, enhancing web-accessibility, and enabling accessibility testing, to mention only a few. Publishers could pretty quickly begin to integrate some of these tools to enhance accessibility of their websites for image recognition, alt text, and quick captions/transcriptions on video content. 

As Helen Simpson, my co-author in Social Media Marketing for Book Publishing says: “While AI can rapidly analyse data and generate ideas, we humans provide the empathy, ethics, values and vision needed to apply that knowledge for the common good.” We should work to our strengths and, within publishing houses, create human-centred approaches to working with AI. This can be as simple as starting with staff, and how they can be supported to do their job in the best, most efficient way possible. In this way, AI can be supportive: not used to replace staff, but to help them free up their time to focus on what is most important.