Bibliographic details and abstracts are available to all. Downloads of full-text dissertations are restricted to University of Portsmouth members who must login. MPhils may be accessed by all.

Abstract

Due to recent developments and interest in Artificial Intelligence (AI), specifically in Large Language Models (LLM), I set up an experiment to analyse the performance of LLMs as a form of video game non-playable character (NPC) against commonly used gaming AI programming techniques. I use a collection of 2D maze environments and technology that allows an LLM to move a character to analyse the capabilities of the AI to correctly move to a specified location and avoid walls. My results help to prove the concept that LLMs could be used to replace traditional game AI technology and provide benefits to run-time calculations and movement flexibility yet also show flaws in logic and calculations within the LLM which require future research and development to identify a solution.

Course: Computer Games Technology - BSc (Hons) - C1671

Date Deposited: 2025-01-10

URI/permalink: https://library.port.ac.uk/dissert/dis14621.html