The boundaries of what’s acceptable

We probably feel a little silly when we do 2+2 in front of a calculator, but we certainly don’t feel like we’re cheating. On the other hand, if we write a prompt to use artificial intelligence, we can already hear the cries of outrage in the air.

The two things aren’t exactly comparable, but they have many elements in common. Both are man-made tools designed to provide concrete assistance in creating something. So let’s start with the concept of an aid and take a broader view, citing a news story from a few weeks ago.

When eBooks learn to speak

On August 31, StreetLib — the distributor we’ve been working with since 2014 — announced a partnership with ElevenLabs. Thanks to this partnership, publishers and authors can distribute eBooks and audiobooks on ElevenReader, an app for iOS, Android, and the web where texts are read aloud using synthetic voices. For audiobooks that have already been produced, listeners are presented with a choice: the original narrator’s voice or a voice generated by artificial intelligence. Claudia Dipilato, StreetLib’s commercial director, stated that this AI is ‘capable of breaking down barriers to access to culture.’

The novel under fire

While synthetic speech doesn’t shock anyone — in fact, it’s held in high regard as a tool for inclusivity — the same cannot be said for texts written using artificial intelligence. Thélyson Orélien, a 38-year-old Canadian writer of Haitian descent, published C’était ça ou mourir, a dramatic and partly autobiographical novel, in Canada in March and in France in August. Immediately hailed by critics, it sold approximately 35,000 copies in five weeks.

Then an anonymous account on X, Balance Ton Claude, claimed that the book had been written almost entirely by artificial intelligence. The evidence, it was claimed, came from an analysis by Pangram, a detection software considered among the most reliable. The massive scandal that ensued led to Orélien’s exclusion from the list of nominees for the prestigious Goncourt Prize. In practice, there is nothing to confirm exactly how the book was actually created and came into being in its final form.

‘No book is written by artificial intelligence’

According to Luca Sofri, this is a fact. On his blog Wittgenstein, the journalist lays the groundwork for a counterintuitive argument. If an author decides to use artificial intelligence software, Sofri says, they do so based on their own ideas and guidance. This has been the case with every significant tool in the history of book publishing — from automated proofreaders to human proofreaders — which for centuries have tempered the notion of absolute authorship. There are only two legitimate grounds for disciplinary action: if someone is guilty of plagiarism or has lied about their work. The problem is that declaring that one has produced something using artificial intelligence goes hand in hand with the perception of subpar work.

The boundaries

So, taking the Orélien case as an example, is the author talented for having known how to use artificial intelligence to the benefit of his own creativity, or must we acknowledge that even that cold, digital entity we invented to help us is endowed with creativity?

Of course, using a calculator isn’t controversial, because a number is always a number, and a calculation yields the same result for everyone. Employing and engaging artificial intelligence for something rooted in creativity, however, is an entirely different story. Are we really reducing contemporary literature to an exercise in style — one that artificial intelligence is certainly capable of performing far better than most writers?

Perhaps we should ask ourselves what the acceptable boundary is within which our creativity can go hand in hand with the use of artificial intelligence, as a tool to help us refine our ideas even further. Accessibility — which has been discussed so much in recent years — ultimately points in this direction for the reader. But what about the creator?

A two-way street

In conclusion, to echo Sofri’s reasoning, expecting those who discuss art to be the sole masters of form is an undemocratic way of excluding from art itself those who have ideas but lack the tools to express them. In daily life and at work, people already rely on artificial intelligence to develop projects, verify information, and even seek psychological help — in an uncontrolled and somewhat blind manner. Since no one is held accountable for what is written, artificial intelligence acts as an oracle. If, on the other hand, someone steps forward and signs their name, the expectation is that everything is genuine.

The solution must go hand in hand with education on how to use the tool. Writers can and should be asked to be transparent about the creative process, unless the use of artificial intelligence is taken for granted. Readers and critics, however, must learn to view it for what it is and not for what it evokes.

For the rest, just like the ElevenReader listener, it is human choice that makes the difference: listening to the artificial voice or still choosing the human one, aware that they are no longer so distinct.