Creative Access Research on AI and Publishing
The Bookseller’s morning newsletter caught my attention with an eye-catching headline, suggesting that publishing has the ‘lowest uptake of AI across creative industries’, according to new research from Creative Access. That’s the kind of statistic that sticks in people’s minds—that might even shape the industry discourse on the topic. It’s also of great professional interest to me given my work delivering AI training and consultancy to more than 330 publishers in 15 countries in recent years. And I have some concerns about how it’s being reported and what people will take away.
The basic problem is that the text in quotation marks in the headline doesn’t actually appear anywhere in the Creative Access report. It’s a paraphrase presented as a quote. The nearest match is a section on page 21 discussing the use of AI specifically in creative workflows. But that’s a narrower claim. The original report carefully differentiates between creative and administrative use of AI. From my own experience and third-party data such as the Book Industry Study Group research last year,1 the latter is where I would expect most publishers to report use, but an overall usage figure isn’t broken out for publishing in the Creative Access research.
The issues around AI in publishing are nuanced and intersectional, and within the bounds of its sample size and methodology, the Creative Access report offers some interesting insights. But I worry that an inaccurate headline may give some publishers the wrong impression of where the industry is, and make a sector that has led the creative industries in training, licensing and practical use appear to be a laggard.
-
Note: the BISG data is not a like-for-like comparison with the Creative Access research: it addresses the same industry, but with a different geographical scope, point in time and methodology. I’m using it here directionally. ↩