Information Equity: Investigating AI Tools’ Ability to Process Searches in Philippine Regional Languages

Abstract

The expansion of Google Translate support to Philippine regional languages, including Bicolano, Hiligaynon, Kapampangan, Pangasinense, and Waray, highlights the growing recognition of linguistic diversity in digital spaces. Despite these advances, AI tools’ effectiveness in handling regional linguistic diversity remains underexplored. Looking into AI’s capabilities to handle less-represented languages, this study investigates AI-driven research tools’ ability to process multilingual queries in these languages, focusing on how tools like ChatGPT, Elicit, and Semantic Scholar retrieve, translate, and interpret search queries. Through qualitative analysis, the study evaluates the effectiveness of these tools in retrieving relevant information and addressing linguistic complexities. Findings reveal that while ChatGPT provides relevant responses, they are often brief, lacking deeper contextual understanding. Elicit, on the other hand, produces accurate literal translations but struggles with irrelevant results, poor abstract summarization, and occasional confabulation. Semantic Scholar presents additional challenges, delivering inconsistent results with accuracy issues, a lack of translation features, and frequent hallucinations. The study emphasizes the need for AI systems to improve their ability to process regional linguistic contextual intricacies, advocating for the development of more inclusive and context-aware AI tools to enhance information access. As champions of information equity, librarians must take a proactive role by staying informed, experimenting with AI tools, and continuously refining their practice, shaping how AI influences knowledge access, ensuring inclusivity, and representation of diverse languages.