Prof. Dr. Christoph Bläsi on the question of where artificial intelligence can help in the book world
The XXIV Mainz Colloquium under the title „Artificial Intelligence in the Book World – Machines as Editors, Machines as Readers?“ on 25 January, 2019 in Mainz suggested, among other things, that the use of appropriate tools can help publishers to increase their efficiency. This prompted questions to Prof. Dr. Christoph Bläsi, who organized the conference.
Q (Christian von Zittwitz, BuchMarkt): A first conclusion after the conference: Are there any areas where artificial intelligence can really be used to increase the efficiency of publishers?
A (Prof. Bläsi): Absolutely – AI can be used, for example, in the initial presorting of manuscripts with a view on topic or target group or in the form of digital user interfaces, where you can simply „consult“ data repositories using natural language and thus also get relevant hits in which the actual search term does not occur at all. The second example shows that, in addition to increasing efficiency, AI can also help to come up with better, more customer-oriented and more customer-friendly products. It gets really exciting, however, when artificial intelligence helps publishers to find new roles in an increasingly complex media environment.
Q: What do you mean by that?
A: We live in times of an abundance of content, especially in the Internet – the role of publishers is therefor likely to develop in the direction of providers of orientation – publishers are not least no longer limited to the ´tree killing business´, as the print-based publishing was sometimes derogatorily called in the 90s ! And if publishers along these lines are „curators“ of content, AI systems never getting tired can help a lot with the selection (from huge amounts of data) and the customer-specific presentation.
A: Unfortunately, this is not the case – dangers of AI on different levels of importance have of course also become clear: learning machines and systems trained with insufficiently selected data can – especially in the media industries – deliver devastating results and if a company relies on the AI components of the large platform providers such as Google, which are often available free of charge via defined interfaces, it helps these platform providers to feed their data leech and, in addition, will be dependent on the fact that the offer in the version once introduced will stay on for a long time – in a potentially ´mission-critical´ application for the company. I wouldn’t want to bet on that…
Q: Were these the only critical aspects discussed?
A: No – we also talked about one of the most fundamental ones, namely that AI systems always tend to strengthen the mainstream, or perhaps better: mainstreams (in the plural), and marginalize phenomena that deviate from this, for whatever reason. And it is at this point that a central task for the book world arises: regardless of the possibly optimized processes and product forms just mentioned, people in the book world are called upon to keep readers awake with unexpected, bulky, witty, original content, to help them blow holes in threatening filter bubbles and echo chambers, as one could say using terms of web criticism!
Q: The publishing industry is currently able to do this even without AI, if I look at the trends in title production in some areas. What did the colloquium bring as a message to the book world then?
A: This observation is of course very true – that’s why the mission mentioned is largely independent from making use of the new techniques on the side of e.g. publishers! This side of this fundamental social mandate, however, a pragmatic recommendation to publishers could be to experiment with „small“ AI applications and, above all, to keep abreast of the developments. Whether such applications carry the „artificial intelligence“ label also used for marketing purposes or simply are clever software is not decisive – „artificial intelligence“ is not a protected term!
Q: Why should publishers do this?
A: Not least because AI can ease the burden of routine tasks in publishing and thus create freedom to concentrate with a stance, experience, practical intelligence and gut feeling on what AI can *not* do adequately and will not be able to for the foreseeable future, either. Such relief does not necessarily have to take place when dealing with content centrally, but also by, for example (if legally possible) intelligently supporting dynamic pricing or simply forwarding e-mails to the publisher automatically to the relevant department.
Q: AI will hardly play a role in book distribution and book selling for the time being, will it?