Evaluating the use of ChatGPT's perceptible affordances to automate affective feedback in gamified platforms
2024 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE credits
Student thesis
Abstract [en]
The aim of the study is to examine how perceptible affordances of ChatGPT 4.0 can facilitate creating automated affective feedback generating artifact which can give feedback specifically tailored to individual players or teams in gamified platforms. The affordances used for the study are ability to integrate with any dataset, ability to input and process large datasets, ability to give a response in relation to a given input and ability to give a quick response. Bisevo was selected for this study due to its suitability, particularly its business processes involving a gamified coaching platform and the potential to automate these processes. Additionally, as a developer currently employed by the company, I possess the capability to design, develop, and test the artifact. At Bisevo the coaches are currently conducting the individual development and feedback discussions to every team or the player either physically or virtually and it has become a challenge to continue with that practice further with the business growth year on year and increased partner companies. Therefore, the necessity to automate the feedback giving process has emerged. The data of five games from three different companies are utilized and integrated with ChatGPT 4.0 in order to identify the exact feed or the prompt engineering together with the question techniques to get more affective feedback through ChatGPT.
The study was conducted following the information system research framework and Design Science Research Methodology (DSRM). The results of the study showed that the affordances such as ability to integrate with any dataset, ability to give a response in relation to a given input and ability to give a quick response can facilitate in creating the artifact. However, there was a limitation in the affordance of ability to input and process large datasets in creating the artifact, even though the input dataset is within the number of tokens and words claimed by OpenAI. Precautions such as getting team-wise analysis and not game-wise is recommended in situations where either the number of teams are higher, or number of behaviours are higher, or both are higher in a given game context. The feed with the selected question technique is to be incorporated in to the coaching platform of the selected organization after testing to create company and ChatGPT integrated, automated affective feedback generating artifact. After testing under different gaming contexts, the artifact can be incorporated into any gamified process where the primary objective is giving feedback and engaging users.
Place, publisher, year, edition, pages
2024. , p. 123
Keywords [en]
Gamification, Affective feedback, ChatGPT perceptible affordances, AI & Machine Learning, Automated feedback, Design science research methodology(DSRM), IS Research Framework
National Category
Computer Sciences
Identifiers
URN: urn:nbn:se:uu:diva-539501OAI: oai:DiVA.org:uu-539501DiVA, id: diva2:1901982
Subject / course
Information Systems
Educational program
Master programme in Information Systems
Presentation
2024-08-22, Ekonomikum, Box 513 Kyrkogårdsgatan 10, 751 20 UPPSALA, 14:15 (English)
Supervisors
Examiners
2024-10-012024-09-302024-10-01Bibliographically approved