CreativeBot: a Creative Storyteller Agent Developed by Leveraging Pre-trained Language Models

Abstract

In an attempt to nurture children's creativity, we developed a creative conversational agent to be used in a collaborative storytelling context with a child. We presented a novel approach to develop creative AI. Our approach uses the four creativity measures: fluency, flexibility, elaboration and originality in order to generate creative behavior. We analyzed and annotated our previously collected storytelling data sets -collected with children-according to our four creativity measures. We then used the extracted and annotated data (636 statements) in order to fine-tune two pre-trained language models (Open AI GPT-3). The two models were aimed at generating creative versus non-creative behavior in a collaborative storytelling scenario. We developed the two models to be able to assess the results and compare them together. We conducted an evaluation to assess stories generated collaboratively between a human and both agents separately (n = 26). Adult Users rated the creativity of the agent according to the stories generated. Results showed that the creative agent was perceived as significantly more creative than the non-creative agent. With the experiment results confirming the validity of our system, we may therefore proceed with testing the effects of the creative behavior of the agent on children's creativity skills.

StoryForge: AI-Powered Narrative with Dynamic Imagery and Voice

“…• Integrate OpenAI's GPT-3.5 Turbo for natural language understanding and text generation. Fine-tune the model on a diverse dataset, emphasizing storytelling contexts to ensure it responds effectively to user prompts [1].…”

“…The scope of the StoryForge project encompasses a comprehensive and user-centric approach to narrative generation, aiming to redefine how stories are crafted and experienced. Firstly, the project will delve into the development and optimization of the AI architecture, leveraging the capabilities of OpenAI's GPT-3.5 Turbo model for natural language understanding [1]. This involves fine-tuning the model to interpret user inputs accurately and generate contextually relevant and engaging story text.…”

“…Challenges in incorporating nonlinear storylines: The linear structure of traditional storytelling can make it difficult to incorporate nonlinear plotlines, flashbacks, or parallel narratives. This can limit the ability to create stories that are more complex, layered, and thought-provoking [1].…”

“…The primary objective of the StoryForge project is to develop an AI-powered narrative generation system that revolutionizes the storytelling experience. By seamlessly integrating OpenAI's GPT-3.5 Turbo model [1], Hugging Face's SDXL for text-to-image generation [4], and Bark for text-to-speech synthesis [10], the system aims to provide users with a comprehensive tool for creating dynamic and engaging stories. The project seeks to empower users by allowing them to input specific constraints, such as genre, character details, key events, mood, and more, guiding the AI in crafting personalized narratives.…”

“…Additionally, the project aims to demonstrate the versatility of AI in creative endeavors, pushing the boundaries of narrative generation by combining advanced language understanding [1], image creation [3], and speech synthesis [5]. By catering to diverse user preferences and providing a platform for the synthesis of text, images, and audio, StoryForge seeks to open new possibilities in content creation and storytelling, showcasing the capabilities of AI to enhance and personalize the creative process [7] [12].…”