AI and the book market
Machines that write novels?

Artificial intelligence is also becoming increasingly important in the book industry. Some hopes are realistic, some more science fiction.

By Matthias Bischoff

In recent years, the further development of artificial intelligence (AI) has led to upheavals in many areas. Particularly in business - and especially in sales and marketing - but also in political decision-making processes, AI has become a powerful tool that we encounter everywhere in everyday life, often invisibly embedded in apps or websites. The topic electrifies many: Springer Wissenschaftsverlag alone lists more than 200 German-language and over 3,500 English-language titles on the subject. But what about the usability of AI in publishing itself - what possibilities are there, what visions for the future?

Artificial intelligence as a trend-trumping PIG?

As is so often the case in this rather traditionally minded industry, future skeptics and progress enthusiasts are at odds with each other: Some fear manipulation of book buyers and readers, while others firmly believe that with the help of AI, their needs can be much better determined and ultimately satisfied in the future. As in all other areas, attempts are being made in the book industry to use AI to emulate human thought and action. With the help of AI, complex tasks are to be solved and, above all, behavior is to be predicted, more precisely: the buying behavior of book buyers. First of all, this involves the sale of books that have already been printed, for which potential buyers can be determined very precisely with the help of algorithms. Anyone who uses websites such as Amazon knows that once you have made a purchase, you will be offered books with similar content every time you visit the site again. Newsletters and messenger services can also respond precisely to the needs of buyers. For Amazon, Google or Facebook, all their customers are gigantic piles of data, transparent and open to all kinds of influences.

But these processes are nothing new by now, and no artificial intelligence is needed, just long-tested algorithmic processes to sell or distribute books in this way. The visions of harnessing artificial intelligence start a step earlier. What books, people in the trade presses around the world are asking, should we publish next year and the year after? The field of literature is naturally left out of the equation, because here readers follow their favorite authors. They are happy about the new books of their favorite writers, whose language they love, almost regardless of the topic they write about.

But below the literary peaks lies the great gray area of genres - historical novels and thrillers, nonfiction and how-to books. For some years now, attempts have been made here to predict the trend topics of the next few years as quickly and accurately as possible. Using meta-data searches, keywords can provide clues as to what is frequently searched for on the web, what shifts are taking place, how many millions of people are browsing certain pages. Conclusions drawn from this enable publishers to find and commission suitable authors for certain topics. Admittedly, all these things are still in their infancy, and people in the industry don't talk about them too openly, since it contradicts the ideal that selection depends solely on an original book idea.

Books write people

Although there are certainly no revolutions to be expected in the coming years, and the cultural pessimists have yet to fear that AI will shape or even take over book production, serious studies conclude that the influence of AI will grow in the coming years. Already in 2019, the Frankfurt Book Fair, together with Gould Finch, published a white paper based on a survey of 233 participants* from the publishing industry in 17 countries. A large majority of respondents expect the importance of AI in publishing to continue to grow. 25 percent of the publishers surveyed have already invested in AI. Nevertheless, they do not expect big leaps: no one wants to rush ahead with large investments so far, since it is by no means certain whether increases in sales can really be realized in the end. Holger Volland, Vice President of the Frankfurt Book Fair, is convinced that publishers expect AI to increase efficiency above all, but also to create opportunities in new business areas, but he adds: "It will be a challenge, however, to integrate new technologies in such a way that the cultural characteristics of the publishing market are taken into account.

The distinctive feature of this market is, after all, human creativity. At the shareholders' meeting of a publishing house listed on the stock exchange, one shareholder asked why the publishing house did not also offer "something like Harry Potter", since money could be made with it. But no AI in the world would think of sending a boy to a magical boarding school and letting him fight a battle for good and evil and his own life. So far, the machine can only recognize what is already there or what is in the air. It can identify trends or popular colors for book covers. But the books are written by people for people.

complex and cost-intensive

It is possible that in a few years, AI will be able to support the work of editors, but not replace it. With the help of AI, it is theoretically possible to scan large quantities of text in a very short time and check whether the manuscripts basically fit into the publishing program. A new software, LiSA (Literature Screening and Analysis), from Hamburg-based QualiFiction, offers analysis of fiction manuscripts based on theme, sentiment, and style, and even determines whether the texts are more lighthearted or somber, complex or simply written. Software like that from QualiFiction, which was named Start-up of the Year in 2019 by the Börsenverein Group of the German Book Trade, will probably establish itself in the publishing industry sooner or later. In the U.S., computer programs are also already being used to evaluate manuscripts, as freelance editor Walter Greulich reported from the future!publish 2019 congress. The basic prerequisite for implementing AI in publishing, however, is the availability of data. However, it is a tremendously time-consuming and cost-intensive process to filter out the relevant information from the many heterogeneous systems - reviews, subscription numbers, cancellations, Amazon reader recommendations, et cetera. For smaller publishers, this is an unfeasible effort.

Nevertheless, in the last few years it has been possible to modernize, even revolutionize, the directory of all available books, which lists around 2.5 million titles, with the help of AI. Here, too, however, the work of industry employees who feed and test the learning system is at the beginning. In the process, a new classification feature has been introduced, known as reading motifs. Complementing product groups and subject classification, they open up new possibilities in the marketing of books by targeting unconscious reading needs such as "excitement," "desire to laugh," or "being surprised." With this new tool, booksellers will be better able to find the right books for their customers in the future.

The potential applications of AI are thus diverse, much is still in its infancy, some will prove practicable, and much will also prove too costly for the industry with its many small and medium-sized market participants. Algorithms can support the work in publishing houses and bookstores, but they will not make people superfluous. The success of a book depends on so many factors that are difficult to calculate that a production of "surefire bestsellers," as some may imagine in their dreams, will remain this for the time being: Science fiction.