Creative Poetry Translation with AI

Hear the latest from the AIDCPT project, sharing the next phase of our work on AI and decolonial translation practices.

Large Language Models (LLMs) are biased, error-prone, and bad for the environment; they generate large amounts of terrible writing, and they are often associated with a simplistic, functional conception of translation. But they can also represent language variety in unprecedentedly fluid ways, and support new kinds of translational creativity. The AIDCPT project has been exploring this potential across many languages, including Miya, Hindi, Greek, Chinese, Arabic, English, Portuguese, Italian and Rioplatense Spanish. Join us to hear about, try out and give your views on our app-in-development which is designed to nurture interesting, inventive and ethically reflective translational interactions with LLMs. 

Dr Deepshikha Behera (The ‘Low’ in ‘Low-Resource’ Languages: Multilinguality and the Politics of Digital Preservation) will examine how state initiatives in multilingual translation and digital preservation privilege certain languages while rendering others invisible. As a result, existing hierarchies of access, power and politics of literary production surrounding marginalised languages are reproduced through state-sponsored digital technologies.

Dr Mary Katherine Newman (Translating La Araucana with AI) will present on her research on translating the early modern Chilean epic, La Araucana (1569-90), using a variety of AIs to consider how they approach the task of translation, focusing on the critical opening stanza.