A ChatGPT-powered learning app that fosters critical thinking and storytelling skills with AI
As generative AI reshapes how stories are created, traditional storytelling approaches fall short of addressing the opportunities and challenges of AI-driven co-creation. How can I develop a learning platform that guides users to explore the creative potentials of AI, while fostering critical reflection of its quality and ethical implications?
Leveraging Bloom's Taxonomy, I break down the learning objectives into the following:
Identify various storytelling elements, such as background, plot, and narrative style.
Organize story ideation and development using structured storytelling framework.
Interpret AI prompting pattern, specifically, how detailed inputs lead to more tailored outputs.
Utilize AI tools for for brainstorming and creative enhancement.
Assess AI-generated content critically using a set of literary criteria.
Reflect on AI's implications for long-term language and creative development.
Insights from AI and storytelling researchers emphasized a robust theoretical framework and adequate scaffolding to promote active learning, guiding me to adopt a progressive learner journey and a storytelling framework based on constructive literary theory.
🔍 Expert Interview Highlight
"AI is a productive content generator, but not necessarily a good storyteller. Consider the inherent factors and elements that make a story compelling."
"Creative writing involves a variety of cognitive processes and skills. Simple instructions or task descriptions are insufficient for engaging the complex learning necessary for these skills."
From user interviews, it became clear that many seek a platform that is customizable and fun to engage with. Accordingly, I prioritized ease of navigation and engaging visuals in interface and interaction design.
🔍 Learner Interview Highlight
"I've always been drawn to storytelling. I am really interested to see how AI can add to my creativities."
"The conceptof integrating AI with storytelling sounds fascinating! But writing an entire story might be too much for me. I prefer something more manageable and less time-consuming."
Combining research with stakeholder feedback, I designed a three-step story-creation journey that guides users to experiment with structured story creation using AI. The following swimlane diagram illustrates how each step facilitates user behaviors and cognitive activities that support the learning outcome.
I built an customzable AI generator using ChatGPT's API, with the front-end interface developed in Webflow, and API calling through Wized.
1️⃣ Developing front-end interface in Webflow
2️⃣ Integrating Open AI's & Airtable's API in Wized
3️⃣ Configuring HTTP request & API response
The front-end and interaction design follows these principles:
📝 Simplicity & fun: Using metaphorical language and concise visual elements, the website creates a fun and engaging story-driven experience.
📝 Structured navigation: Content is organized under distinct, clickable drop-down sections that provide clear structure and easy navigation.
📝 Easy editing: Custom codesare embedded to transfer user inputs across sections, reducing repetition and extraneous cognitive load.
The platform was tested with 7 adult participants at its final showcase at Teachers College, Columbia University. Many reported that the interactions offered them divergent inspirations for enhancing their own storytelling, as well as reflections on AI's constraints, such as the rigidity and formulaic patterns in its generation.
🔍 Learner Feedback Highlight
"The systematic breakdown of the story elements is really helpful. I’ve never planned story writing so strategically before."
"The variety of AI-generated stories is impressive. They stimulate a lot of thoughts in me on how to creatively retell the story!”
“After several trials, you start to notice certain patterns in its content and it becomes repetitive."
Two main areas for future improvement:
✅Information breakdown and interactivity: Creative writing is cognitively demanding. Many participants rely heavily on the generator without thoughtfully creating their own narratives first, which prevents meaningful comparisons and reflections. Future versions might include: 1) presenting the information through adaptive module-based tutorials for easier assimilation 2) including gamification elements, such as personifying the AI generator to enhance user engagement, and 3) incorporating AI-driven recommendation and feedback throughout story development process to guide and inspire users.
✅ Conversational AI interaction: The story structure limited user's AI input to predetermined story elements, resulting in restricted text refinement. Many users expressed a desire for more flexible AI interactions, similar to the conversational experience with ChatGPT. Future iterations could incorporate AI into a conversational interface or Avatar, allowing more nuanced control over content.
In the first phase, users are guided to ideate their story components using a narrative framework ("Personal Magic Recipe”), which breaks down the story into the background (“The Story's World”), the plot (“The Plot”), and the narrative style (“The Art of Narration”). Using this framework, users will create an initial story without AI assistance.
This step fosters understanding and application of structured storytelling.
Then, users will recreate their stories using the AI generator embedded into the website. Three versions will be generated, each adding more detail and complexity to the previous. They'll then compare how the tweaks lead to different results in output, and build an understanding of AI prompting patterns through iterative and incremental experimentation.
In the first round ("Providing basic information"), users instruct the AI generator to produce a story with minimal plot, setting, and character information.
In the second round ("Adding depth to your settings and characters"), they will put in additional setting and character details.
In the last round ("Specifying the narration"), users further specify stylistic preferences such as genre, style, tone, etc.
Throughout the different rounds, users have the flexibility to modify the inputs until they achieve their desired creative vision.
In the last stage, the users are guided to critically compare human and AI-assisted versions using 7 assessment criteria, reflecting on the strengths and weaknesses of AI-generated stories and how they would conceptualize their strategies differently. Then, they will have the opportunity to refine their narratives by integrating the strengths of both human and machine intelligence.
This exercise aims to foster reflection and critical thinking on AI content generation.
After completing all these steps, users are prompted to reflect on the entire learning experience for ongoing learning and development.