Minority Language Wikipedias in an AI-Dominated World
Date:
Keynote address.
Abstract: Large language models have enabled the development of advanced AI tools like ChatGPT, and have also greatly improved the quality of machine translation engines over the last 5-10 years. But these models are “data hungry” in the sense that many billions of words of high-quality text are required to get reasonable results. I’ll talk about the state of affairs in the development of these resources for Irish and other minority language communities, and their potential impact, both good and bad, on our Wikipedias.
- Slides: celticknot24.pdf