@ve: A Chatbot for Latin

TL;DR

This paper introduces @ve, a Latin chatbot built with GPT-3.0 and a knowledge base, aiming to aid language preservation and education through interactive conversation, despite current limitations.

Contribution

It presents the development, implementation, and evaluation of a Latin chatbot using GPT-3.0, highlighting its potential for language preservation and teaching.

Findings

The chatbot can communicate in Latin with a knowledgeable base.

Expert testing shows potential but reveals glitches and limitations.

Future improvements include using GPT-4 and expanding the knowledge base.

Abstract

Dead, extinct, and endangered languages have been preserved primarily through audio conservation and the collection and digitization of scripts and have been promoted through targeted language acquisition efforts. Another possibility would be to build conversational agents that can master these languages. This would provide an artificial, active conversational partner which has knowledge of the vocabulary and grammar, and one learns with it in a different way. The chatbot @ve, with which one can communicate in Latin, was developed in 2022/2023 based on GPT-3.0. It was additionally equipped with a manually created knowledge base. After conceptual groundwork, this paper presents the preparation and implementation of the project. In addition, it summarizes the test that a Latin expert conducted with the chatbot. A critical discussion elaborates advantages and disadvantages. @ve could be a new tool for teaching Latin in a memorable and entertaining way through dialogue. However, the present implementation is still too prone to glitches for stand-alone use - i.e., without the accompaniment of a teacher. The use of GPT-4 could be a solution as well as the extension of the knowledge base. In conclusion, it can be argued that conversational agents are an innovative approach to promoting and preserving languages.