The next ten years: weighing the costs and benefits of using artificial intelligence in the Australian publishing industry
Vienna Daniels
Artificial Intelligence (AI) has inspired a medley of anxiety, excitement and curiosity within the Australian publishing industry in recent years. As the technology rapidly develops, many are beginning to question the role that AI will play within publishing, and whether this change is positive or threatening. This paper will endeavour to assess the current state of AI technology, its possible benefits for Australian publishing and whether those benefits outweigh any potential costs to answer the question of whether AI should be embraced by the industry, or if publishers should be concerned.
The current state of affairs
The Artificial Intelligence industry is experiencing rapid growth of unprecedented levels. Now more than ever, companies across all industries are beginning to experiment with emerging technologies in the hopes of improving productivity, profit and growth. Before exploring a possible relationship between AI and the Australian publishing industry, we must first analyse how the technology is being used by adjacent industries in the current day.
Previously guarded behind the closed doors of Big Tech, AI software developers have opened access to the general public for experimentation and implementation across a wide range of applications. Research laboratory OpenAI has spearheaded the AI revolution with their GPT system. Its most recent iteration, GPT-4, has brought to the arena an experience that is far more responsive, intuitive and intelligent than its predecessors. Tech companies have been quick to take advantage of GPT-4 (and the previous GPT-3), integrating the software into their own products. OpenAI’s website boasts a multitude of partnerships, including Duolingo, Khan Academy, Microsoft (Bing) and the Icelandic government.
A slew of other corporations have joined the race with their own AI powered improvements on their pre-existing products, particularly within creative industries. Adobe have released an AI image generator, Canva now allows users to edit their own images with prompts and cloud-based typing assistant Grammarly has streamlined their service with AI assistance. Many of these innovations are user-focused, designed to make otherwise complex processes more accessible and streamlined. These companies are using AI to create an all-in-one experience, preventing the user from needing to look elsewhere for certain services such as editing, graphic design and transcription.
But how are businesses using AI internally? Process automation is where AI really shines. Research carried out by the Harvard Business Review found that the most common area within which companies utilised AI was in automating a series of administrative tasks that would typically be carried out by humans. Robotic Process Automation (RPA) denotes software that uses robots to compute and quickly perform processes. It is a more basic form of AI that doesn’t evolve through machine learning and will work within parameters set by the company that is implementing it. The Harvard Business Review identified RPA as ‘the least expensive and easiest to implement of [other] cognitive technologies,’ concluding that ‘if you can outsource a task, you can probably automate it’. Keep in mind, this data is from 2018—in the last five years, RPA has become an even more prominent part of how corporations function in the day-to-day. As the technology improves, the range of applications begins to expand and as do the complexity of responsibilities that RPA can perform.
Another significant internal application of AI is within the decision-making process. Research from the MIT Sloan Management Review discovered that ‘it’s possible for companies today to automate entire decision-making processes and autonomously substitute human involvement in the final choice through AI integration’. The Review found that while companies are using AI to accelerate the analysis of data and generate subsequent recommendations, the final decisions made still fell on the shoulders of executives. The conclusions these executives drew from the AI generated data varied significantly from individual to individual, highlighting AI as a helpful tool but not yet an essential factor to the choices made.
Given that AI development has accelerated so significantly in the past few years, we are still yet to see the true potential of the technology, and thus its real-world applications. Still, there are many opportunities for businesses of all shapes and sizes to adopt the programs and utilise them in their day-to-day operations.
A snapshot of the Australian publishing industry
The publishing industry is currently in a state of metamorphosis. Recuperating after the Covid-19 pandemic has been an uphill battle—Australia experienced somewhat of an [uptake in physical book sales][8] during the nation’s long lockdowns, but in the years since we have returned to ‘normal’, [revenue has dipped below pre-2018 numbers][9]. Australian publishers and booksellers are also facing intense competition from online retailers that are based outside of the country. As [IBISWorld reports][10], ‘Australian publishers have been negatively affected by the popularity of overseas-based online retailers, such as Book Depository, which allow consumers to purchase foreign-published books directly from their websites’. Increased adoption of online purchasing during the lockdown-era has driven buyers away from local sellers and toward major online markets, such as Amazon and AbeBooks.
[10]: Australian publishers have been negatively affected by the popularity of overseas-based online retailers, such as Book Depository, which allow consumers to purchase foreign-published books directly from their websites [9]: https://www.ibisworld.com/au/industry/book-publishing/171/ [8]: https://publishers.asn.au/Web/Web/Latest/APANews/Publishing-industry-discusses-the-challenges-of-COVID-19.aspx
Educational publishing
With regard to educational publishing, more schools and universities are prioritising digital access for their textbooks. Having a digital version of your textbook is now a non-negotiable for educational publishers, and trying to navigate the digital world while still maintaining a profit is proving difficult. Protecting digital textbooks from illegal sharing has prompted educational publishers to build their own digital platforms or share a portion of their profits with a retailer such as Campion that consolidates digital resources from multiple publishers in one place for schools and students to access. It hurts to share a slice of the profits but the practice has become a necessary evil for educational publishers to remain relevant in the Australian industry and ensure adoption of their textbooks.
Supply chains
It’s no secret that Covid-19 caused major disruptions in the supply chains across all industries. Freight costs have increased to unprecedented rates and have failed to recover in the post-Covid era. Local Australian printers appear to be shutting down one after another, with even locally sourced printing supplies becoming scarce—Australia’s only commercial paper mill closed its doors in December of 2022. Australian publishers are forced to rely on imports to deliver their physical texts but repeat supply chains issues and increased costs have put the role of demand planning to an even higher importance. Demand planners need to be able to keep up with the evolving industry and unpredictable supply chain interruptions, while keeping costs low and avoiding waste. Miscalculating demand could mean sellers are left waiting for months for stock to arrive from overseas printers and publishers lose out on sale opportunities.
Children and book consumption
Fewer children are reading now than ever before. A survey carried out by the Australian Bureau of Statistics found that only 72% of children aged between five and fourteen read for pleasure, with most reading for two or less hours per week. In stark contrast, 90% participated in screen-based activities and 24% of those children used screens for more than twenty hours. Publishers are faced with the challenge of accessing a generation that is increasingly reliant on screens, both at school and at home. This has led to publishers moving away from just standard print books and toward creating alternative digital content to either replace print or to accompany their physical catalogues. There is a big problem in attempting to break into the digital market—the online sphere is already inundated with free resources across all platforms, including websites, blogs, YouTube and other social media sites. As the Copyright Agency noted back in 2019, ‘in a world flooded by often mediocre, if not dubious digital content across all media platforms: film, television, radio, news sites, blogs and music, to name but some, the educational world too is awash with content volume rather than content quality–and like the truth–quality is very often the first casualty’. Creating quality content for children is a costly process that involves expertise, verification and skilful presentation, but the sphere is already dominated by content creators and family-oriented influencers.
Sustainability
Sustainability has been a concern for Australian publishers for quite some time, but there are still a barrage of challenges preventing houses from being as eco-friendly as they can be. Most publishers work with sellers on a Sale or Return basis (SOR). SOR practices allow booksellers to return stock that does not sell to the publisher, typically within a certain time frame and up to a certain percentage of their total purchases. The nature of publishing means that there is always a certain amount of risk associated when it comes to selling books—publishers can’t be certain as to which books will be hits and which will fail to sell, meaning that there is potential for unsold texts to be sent back to publishers in droves. The constant shipping back and forth of books is an ecological nightmare. In such a large country, books must travel far between warehouses and shops, with transport comprising almost a quarter of greenhouse gas emissions in Australia. There is significant wastage associated with damages occurring during transport. If a book has been damaged, many publishers require it be returned to them or destroyed by the sellers, therefore making it unable to be sold or donated to charity. Damaged books collectively worth thousands of dollars are wasted every year, with many of the damages being minor and not necessarily rendering the text unusable.
The range of problems facing the Australian publishing industry are wide and complex, with no easy or straightforward solutions. The constantly evolving nature of the industry, combined with strong competitors rising from overseas, has put Australian publishers in dire straits. Covid-19 threw an unprecedented spanner in the works with long lasting implications felt at all levels of the supply chain. Confronting these challenges requires innovation, an open mind toward emerging technologies and a willingness to embrace the unknown⎯a challenge for an industry that has been struggling to stay afloat in the modern era.
Utilising AI in the Australian publishing industry
No matter how large a house grows, customer service remains a constant priority. The term ‘customers’ includes booksellers, direct consumers and users of digital content, both paying and otherwise. As tech literacy grows, many now choose to use computers and text-based interactions with customer support teams over calling them on the phone. This allows for many opportunities in which AI based technology can be used to expedite interactions with customers and deliver satisfactory service without the need for human intervention.
Conversational AI
Conversational AI has developed at a rapid pace, making interactions smoother, more natural and more helpful with the assistance of natural language processing (NLP). NLP is what has helped recent AI fads, such as ChatGPT, to be so impressive⎯these conversational models now make ‘talking’ to an AI powered robot far more intuitive and straightforward compared to their predecessors. These AI chatbots are especially helpful given that they can be available 24/7, allowing customers to access assistance whenever they need it, as opposed to only during traditional working hours. Furthermore, utilising a chatbot to handle more basic customer service tasks will allow a publisher’s customer support team to spend more time on complicated queries and tasks, raising the level of productivity within the team as a whole. Both customers and customer support employees will see the benefits here.
Business processes
In terms of capitalising on AI powered technology internally, businesses have a myriad of options available. In its current state, AI provides the most use within automating various administrative processes that are currently done manually within publishing houses by using RPA.
Purchase orders (POs) are the heartbeat of publishing in Australia⎯booksellers, organisations and more submit their POs to publishers and they are then manually inputted into their data processing application of choice and outputted to their warehouse for dispatching. POs can come in many different shapes or sizes, with various amounts of information provided and a wide range of formats. Human input is used to decipher the important information such as the ISBN ordered, the quantity and the address that the items are to be shipped to. Currently, it is a laborious, time-eating practice that publishers must endure. The use of Intelligent Document Processing (IDP) software can extract any necessary information from a PO and submit it directly to the warehouse, speeding up the ordering process and ensuring that books are in the hands of sellers more quickly.
Data analysis
Quick data analysis and output generation, such as the tech seen in IDPs, can also be of great use in other parts of the publishing supply chain. Demand planning is likely to see great benefits with algorithmic assistance from AI. AI tools can draw data from previous sales, the current publishing landscape, emerging trends and hard numbers such as current stock levels, pricing and more to crunch the data and produce more accurate predictions, more quickly. This processing power is also key to more sustainable publishing. As Alex Koshulko from Forbes notes, ‘AI solutions can constantly rebalance demand and supply by automatically analyzing all available data and constraints’. Having constantly up to date data and subsequent analysis will help to prevent publishers from overstocking on certain titles, ensuring there is less wastage and fewer emissions produced from SOR claims.
Future possibilities
Director of Sales and Marketing at Oxford University Press Australia and New Zealand (OUP ANZ) Andrew Ritchie was asked for comment about the current place of AI within Australian publishing, and what possibilities might present themselves in the future. He noted that while OUP ANZ is not currently employing AI in its everyday processes, there is certainly an interest in the discussions on how the publisher might start utilising technology. One idea we discussed was the possibility of training AI models on a centralised database of Oxford’s educational content: ‘Certain topics and content is specific to the region that we’re producing for, but other subjects like algebra are pretty much the same everywhere. We could somehow create a database of this information and then prompt an AI to create a textbook, for example, that uses this data and modifies it for the target audience.’ The idea would be that this technology could remove some of the more monotonous work of essentially rewriting information that already exists within OUP’s catalogue elsewhere.
When looking at currently available technology, there are already an abundance of practical applications that publishers can benefit from. Considering how AI will continue to evolve⎯and the rapid pace at which it has developed over the last few years⎯there is this notion that the sky is truly the limit. The hard part is figuring out just how this technology can be used specifically for the Australian industry, which has its own set of unique challenges to deal with. Furthermore, imagining a future in which AI is heavily incorporated within the industry comes with its fair share of concerns regarding job security, a loss of creativity and the monetary costs associated with overhauling the industry’s current processes.
The costs of using AI in the Australian publishing industry
The big question on everyone’s mind is whether AI powered technology is going to make certain roles within publishing redundant. These concerns are not completely misplaced, as recent research from Goldman Sachs suggests that 18% of all work around the globe could be automated using current generative AI technology. Industries requiring manual labour and human-to-human interaction⎯such as agriculture and healthcare⎯are least likely to be affected, while those involving data analysis, content creation and calculation will be the first to be impacted. Where does that put publishing? As explored earlier, publishing is ripe for AI incorporation given its high volume of administrative work, from copyediting to demand planning. In its current state, AI will not be able to carry out the high majority of the work that humans are completing but, considering the rapid pace of develop ent we have been witnessing, that may not be true in the coming years.
Copyright concerns are a hot topic in the AI discourse, particularly when it comes to generative content, since AI models are trained on pre-existing information and data. When used in a creative space, such as to create art or text, this forms a grey area around who owns what content, and how it can be used legally. AI is developing at a rate far quicker than our legislation can keep up. Pre-existing copyright laws have not taken into consideration machine generated works, and no precedent has been set to guide the use of AI in creative industries. As the Arts Law Centre of Australia notes, ‘it is currently not clear whether [copyright] exceptions cover use of works as training data in machine learning. They were not designed to apply to that type of use. There are currently no copyright exceptions in Australia specific to data mining or using works for machine learning’. Employing the use of AI in published content could mean a copyright nightmare for publishers.
Lastly, completely overhauling current practices in an industry is no easy feat. The potential applications described in this paper will require time, money and effort to implement. For smaller houses, these changes might be easier to add to their processes, but for larger, global publishers the operations will be much more complex. It is also important to note that not everyone will be willing to embrace these changes. Tech literacy can vary from person to person and, depending on the individual, there may be some resistance to embracing this new technology. Convincing a company to invest in new technology is one thing; making everyone get on board with it is another. These are both exciting and terrifying times, and taking a risk on AI might be too radical for some.
What happens next?
If there is anything certain in regards to AI technology and the Australian publishing industry, it is this: AI is coming whether we like it or not, and the publishers that fail to effectively incorporate AI into their processes are going to get left behind. Like any major evolution in technology, it is unlikely to be a smooth transition, but in ten years there will be two distinct camps of publishers: those who failed to utilise AI as well as they could, and those who did. The ideas presented in this paper are just the tip of the iceberg when it comes to practical and effective methods with which AI can be applied to the Australian industry today. While there is likely to be some significant costs associated with this new era, and some pushback from those affected, the benefits are too great to be ignored. AI can offer practical solutions to some of the biggest challenges facing Australian publishers and to not at least explore these options would be misguided. AI does not have to be an existential threat but rather, when used correctly, an invaluable tool to ensure both the survival and the prosperity of the industry.
About the author
Vienna Daniels is a hopeful publisher who has spent twenty-something years trying to conquer the English language. After a four year stint in Canberra, Vienna returned to her home of Melbourne to further this goal - despite an unchanging abhorrence toward rainy weather. She hopes a Master of Writing and Publishing from RMIT will get her one step closer, though this remains to be seen.