When AI is not the solution: from principles to practice

Abstract

This talk will present the research process to develop a conversational agent to support and dignify the work of indigenous languages interpreters in Mexico. The project was an exploration to co-design the system and the research methodology with the end users (indigenous interpreters) in a way which aligned with the principles of: gender perspective, collaboration, shared benefits, digital autonomy and data sovereignty. The talk will explore the political, social and cultural landscape of indigenous languages interpretation in Mexico, along with key aspects of indigenous data sovereignty, to provide contextual basis for the project and its proposed methodology. Finally, we will explore the social conditions which led to the decision that AI was not the right tool to tackle the problem, if the aim was to develop a technology which in all its stages, from its design to its deployment, aligned with set of principles which were chosen.