Another day passes and with it the seemingly unstoppable creep of AI and the automation of our sector.

Last week, Amazon-owned Audible announced plans to launch a new AI tool which would give publishers and authors the option to automate the production of audiobooks for use on the platform. Slipped in alongside this announcement was also the news that Audible would soon be rolling out automated translation services.

Audible presents this as a big step towards the democratisation of audiobooks, a rapidly growing format. Authors and publishers will, in theory, be able to reach wider audiences, markets and cultures at a fraction of the cost, regardless of their profile or name recognition. A huge win for the writing community, surely?

Sadly not. The automation of the audiobook production process may present some benefits, but the automation of translation should concern us all. It would be shortsighted and unrealistic to suggest that AI should have no place in literary translation or the publishing industry in general. After all, we’ve used spellcheck on this piece and wouldn’t expect others to forgo it. However, the treatment of literary translation should be observed closely and carefully.

Translators are the canaries in the coal mine. What happens to them will ultimately reflect the future trajectory of the wider literary sector. If Audible can replace human translators and narrators in a heartbeat, then it is not implausible to conclude that other people and parts of the creative process could be replaced too.

Translating a published work is so much more than running a text through DeepL Translate. It is a challenging and fascinating art form. Entire new interpretations, conclusions and layers of nuance can be discerned from how a body of work is translated. It is why certain bodies of work have multiple different translations.

Emily Wilson’s translation of Homer’s The Odyssey received widespread attention and acclaim, despite being the 14th English translation in the 21st century, for example.

Our fear is that this pursuit of democratisation by AI will make automated and synthetic translation the norm, and human translation a luxury reserved only for writers with the highest of profiles

And while AI-generated translation may lower the financial costs and the amount of time it takes, there is a real danger that the process could drive readers away from translated works. Just think, would the runaway success of Asako Yuzuki’s Butter have occurred if it had been translated by Audible rather than the incredible Polly Barton? In the quest to use automation to introduce authors to new audiences, we run the risk of actually deterring readers from translated works entirely and needlessly sacrificing the art and profession of literary translation in the process.

The same thing is true beyond our own shores. How can we expect international audiences to appreciate the works of UK authors if our great British wit and subtext are clinically stripped away by an algorithm? Audible recognises this problem too, assuring potential users they will be able to request that a professional linguist review the output “to ensure translation accuracy and cultural nuance”. This strikes us as a deeply flawed process.

There is, of course, a case to be made that this technology will open up translation to smaller authors, thereby granting opportunities not afforded to them in the past. However, our fear is that this pursuit of democratisation by AI will make automated and synthetic translation the norm, and human translation a luxury reserved only for writers with the highest of profiles. Such a development does little for fledgling and niche writers and could widen the existing disparity between writers with large and small profiles.

If automated translation becomes the norm, what is to stop the same from happening to other essential aspects of publishing? Illustration? Screenwriting? Playwriting? Entire novels drafted by AI could become the standard, while human-authored works are reserved for established writers and celebrities who are guaranteed to draw an audience. Clearly, this cannot be allowed to happen. Automated translation is here to stay, but that doesn’t mean that it must become a prevailing and dominant aspect of the publishing process.

First and foremost, there must be transparency at every stage of the process, between authors and their publishers, as well as between consumers and booksellers. Writers must have a choice as to whether their works are translated by AI, a human or not at all – not publishers. Similarly, works that have been artificially translated must be clearly signposted so that readers are aware that they are consuming an artificially translated text.

In our view, this strikes a fair balance between preserving the interests of writers and translators and embracing the opportunities that new technologies present to the industry. Indeed, writers have a long history of embracing and adopting new technologies and innovations to support their creative process.

Writers and translators have faced an increasingly challenging environment for over two decades, and this has only intensified in the past five years. In this context, it is crucial that the creator community, including authors and translators, is not shut out from the decisions behind the application of AI in publishing. If AI can lower costs and create opportunities, the benefits of this should be shared with writers, and not hoarded by publishers and booksellers.

Automated translation could be a valuable lifeline to some authors where professional literary translation is not feasible. However, we are united in our belief that it absolutely must not become the norm. To quote the award-winning translator Frank Wynne, “adequate is a poor substitute for human”.