In a detailed presentation, Schoppert took viewers back to the beginning of AI tools, through to the evolution of transformers, before looking at how the current models work and their modular character. 

He then turned his attention to the kinds of datasets being used and amount of (unlicensed) books being used to train the current tools. He considered the legal context of this situation from both fair-use and copyright exception perspectives as well as looking at consideration about moral rights of attribution among other issues. 

The webinar closed with a dynamic Q&A session where Peter Schoppert and Jessica Sanger gave their perspectives about key issues around the impacts of AI in publishing and on prominent copyright related questions.

IPA members, and their members, can watch a recording of the session upon request [UPDATE 5 June] can watch the webinar on the IPA Academy. Contact your Publishers Association for your access code.  You can follow Peter Schoppert’s blog posts on AI at 

While Peter and Jessica’s contributions were their own personal views about AI, IPA is closely monitoring AI policy and law related developments around the world and has intervened in several public consultations on this important matter. We will continue to inform and consult members.