Reflections on Cope and Kalantzis and the intrinsic multimodality of generative AI

In Cope and Kalantzis's recent discussion in this journal concerning the organization and capabilities of Generative AI, they argue that such systems exhibit critical limitations in their ability to engage with meaning and multimodality because of an alleged deep foundation in digitized ‘written texts’. In response to this, I explain why the core components of Generative AI are inherently multimodal and are not restricted to written language, or any specific modality, at all. This supports their use in an ever broadening range of multimodal analysis contexts, of which I give several examples. I then point to a more general semiotic position from which to critically reflect on their capabilities in the future.