Paula Helm articulates an AI vision that goes beyond base performance to include epistemic justice and cultural diversity by focusing on speakers and not language alone.

Paula and Kimberly discuss ethics as a science; language as a core element of culture; going beyond superficial diversity; epistemic justice and valuing other’s knowledge; the translation fallacy; indigenous languages as oral goods; centering speakers and communities; linguistic autonomy and economic participation; the Māori view on data ownership; the role of data subjects; enabling cultural understanding, self-determination and expression; the limits of synthetic data; ethical issues as power asymmetries; and reflecting on what AI mirrors back to us. 

Paula Helm is an Assistant Professor of Empirical Ethics and Data Science at the University of Amsterdam. Her work sits at the intersection of STS, Media Studies and Ethics. In 2022 Paula was recognized as one of the 100 Most Brilliant Women in AI-Ethics.

Related Resources

• Generating Reality and Silencing Debate: Synthetic Data as Discursive Device (paper) https://journals.sagepub.com/doi/full/10.1177/20539517241249447
• Diversity and Language Technology (paper): https://link.springer.com/article/10.1007/s10676-023-09742-6

A transcript of this episode is here.