Examining the applicability and attitudes towards artificial intelligence as a research assistant

Abstract

ChatGPT has proven to be a breakthrough in the field of artificial intelligence since its release to the public in 2022. With the popularization of large language models and generative artificial intelligence tools, it is inevitable that their use has become both a topic of discussion and a challenge for academia. This study examines the applicability of using a chatbot as a research assistant when trained with only a small number of relevant articles, using Azure’s OpenAI chatbot environment, and the general sentiment researchers have towards the use of artificial intelligence in academia. The thesis aims to determine what tasks researchers prefer to use chatbots as an assistant for, what tasks the technology is generally capable of performing successfully, and the possible broader implications and effects of these tasks being performed by artificial intelligence.

Using the Design Science Research Process, a chatbot artifact was designed, trained and evaluated during this study. The results of the study support the trends shown by past research, both in applicability and sentiment. The artifact chatbot proved to be capable of assisting in some research tasks, but not to the extent where it is suitable for academic use without human supervision. The general sentiment towards the use of the artifact was positive but showed variance both depending on the researchers and the intended purpose of the chatbot.

The target demographic for the artifact was a research unit in the University of Oulu. Many of the participating researchers showed interest in the application of artificial intelligence in academia. The thesis examined how a chatbot tool could benefit small demographic of academics and shed light on their opinions regarding the broader usage of chatbot tools.

This study reinforces the findings of past research and highlights possible angles for future research regarding the topic in academia and beyond. As such, this thesis contributes to the vast amount of literature regarding the topic of chatbots, ChatGPT and artificial intelligence. The results of this study both support the notion that the technology must be tackled as a subject in academia and serve as a possible source for new ideas regarding the safe and useful application of artificial intelligence tools for the broader academic community.