Hatevolution: What Static Benchmarks Don't Tell Us
This addresses the reliability of model safety assessments for hate speech detection, highlighting an incremental need for time-sensitive benchmarks.
The paper tackled the problem of static benchmarks failing to capture the evolution of hate speech over time, showing that 20 language models exhibit temporal misalignment in evaluations, with concrete evidence from two experiments.
Language changes over time, including in the hate speech domain, which evolves quickly following social dynamics and cultural shifts. While NLP research has investigated the impact of language evolution on model training and has proposed several solutions for it, its impact on model benchmarking remains under-explored. Yet, hate speech benchmarks play a crucial role to ensure model safety. In this paper, we empirically evaluate the robustness of 20 language models across two evolving hate speech experiments, and we show the temporal misalignment between static and time-sensitive evaluations. Our findings call for time-sensitive linguistic benchmarks in order to correctly and reliably evaluate language models in the hate speech domain.