Generating Role-Playing Game Quests With GPT Language Models

Quests represent an integral part of role-playing games (RPGs). While evocative, narrative-rich quests are still mostly hand-authored, player demands towards more and richer game content, as well business requirements for continuous engagement necessitate alternative, procedural quest generation methods. existing methods produce uninteresting, mechanical descriptions, recent advances in AI have brought forth generative language models with promising computational storytelling capabilities. We leverage two the most successful Transformer models, GPT-2 GPT-3, to procedurally generate RPG video descriptions. gathered, processed openly published a data set 978 their descriptions from six RPGs. fine-tuned on this range optimizations informed by several mini studies. validated resulting Quest-GPT-2 model via online user study involving 349 players. Our results indicate that one five would be deemed acceptable human critic, yet variation quality across individual is large. provide recommendations current applications Quest-GPT-2. This complemented case-studies GPT-3 highlight future potential state-of-the-art natural generation.