The Hrunting of AI: Where and How to Improve English Dialectal Fairness
This work addresses fairness and inclusivity issues in AI for underrepresented English dialect speakers, though it is incremental as it builds on known data scarcity problems.
The study investigated the challenge of improving large language models (LLMs) for English dialects, finding that low human-human agreement on quality in low-population dialects (e.g., Yorkshire, Geordie, Cornish) directly limits LLM performance and alignment, with fine-tuning potentially worsening this pattern, but identified some LLMs' ability to generate high-quality data as a scalable solution.
It is known that large language models (LLMs) underperform in English dialects, and that improving them is difficult due to data scarcity. In this work we investigate how quality and availability impact the feasibility of improving LLMs in this context. For this, we evaluate three rarely-studied English dialects (Yorkshire, Geordie, and Cornish), plus African-American Vernacular English, and West Frisian as control. We find that human-human agreement when determining LLM generation quality directly impacts LLM-as-a-judge performance. That is, LLM-human agreement mimics the human-human agreement pattern, and so do metrics such as accuracy. It is an issue because LLM-human agreement measures an LLM's alignment with the human consensus; and hence raises questions about the feasibility of improving LLM performance in locales where low populations induce low agreement. We also note that fine-tuning does not eradicate, and might amplify, this pattern in English dialects. But also find encouraging signals, such as some LLMs' ability to generate high-quality data, thus enabling scalability. We argue that data must be carefully evaluated to ensure fair and inclusive LLM improvement; and, in the presence of scarcity, new tools are needed to handle the pattern found.