Measuring Harmful Representations in Scandinavian Language Models

TL;DR

This study investigates gender-based harmful and toxic content in Scandinavian language models, revealing they contain stereotypes despite the region's reputation for gender equality, highlighting potential risks in real-world applications.

Contribution

The paper introduces a systematic probing method to evaluate harmful content in Scandinavian language models across Danish, Swedish, and Norwegian.

Findings

Models contain gender stereotypes similar across languages

Harmful content exists despite regional gender equality reputation

Results highlight risks of deploying these models in practice

Abstract

Scandinavian countries are perceived as role-models when it comes to gender equality. With the advent of pre-trained language models and their widespread usage, we investigate to what extent gender-based harmful and toxic content exist in selected Scandinavian language models. We examine nine models, covering Danish, Swedish, and Norwegian, by manually creating template-based sentences and probing the models for completion. We evaluate the completions using two methods for measuring harmful and toxic completions and provide a thorough analysis of the results. We show that Scandinavian pre-trained language models contain harmful and gender-based stereotypes with similar values across all languages. This finding goes against the general expectations related to gender equality in Scandinavian countries and shows the possible problematic outcomes of using such models in real-world settings.