Dragon’s Delusion

Cheng Xu, KongKee, Lawman, Hanjun Dai, Yixin Wang

Dragon’s Delusion is an AI-powered storytelling tool that invites users to share their stories and helps workshop them, one question at a time. Sharing stories is powerful. Our mission is to democratize storytelling and history-making, giving voice to the underrepresented and empowering them to take control of their own narrative.

Concept

Our team is fueled by a vision of digital democracy where history is no longer shaped by the hands of a privileged few.

We had a very thorough survey of the ever growing set of AI generative tools. Current AI design tools generally take on the role of a creative lead. In storytelling tools, AI would generate characters, relationships, and plots. The intention is to get anyone, with or without writing expertise, to complete a story. By doing so, we might risk putting the tool in a superior position than a novel user. Why is that?

Before AI design tools gained popularity, other design tools have taken on similar roles of expanding user intention with computational ‘magic’. In one tool I observed that the augmentation of the user’s ability brings a sense of joy, but also leaves some users with the impression that the computer is better at designing. The confidence in one’s own design ability remains unchanged.

Having surveyed the AI generative tools, we want to position AI as a partner that leaves the user as the leader, and plays the role of an encouraging partner. We believe that the intention and stories are already in everyone. They just need a space and a trusting space to talk through them. 

Our initial goal was to learn through deep collaboration with artists and technologists in order to create a tool that is accessible for the general public. The team met online and offline regularly to find the right problem to tackle. We also ran a six week program with Edge on Square to test out some AI generation tools. 

Our problem selection criteria included: 

1. Should it be done: What unique historical, artistic, or technical context makes the idea compelling?  Does it have to be now or can it wait?  

2. Can it be done: Is this achievable with the current tools?   

3. Are we the ones to do it: Is the team positioned to tackle part of the problem in the duration of the C/Change Lab?

Process

We went through many possibilities, and one story from our team member Lawman stuck with us. When teaching in Hong Kong, he had the class interview their parents about how they fled into Hong Kong. That opened up unheard family stories and deepened generational connections. If that kind of connection could be replicated with our tool online, we would be able to form communities and write history directly with people whose stories are typically erased. In a time of war and media suppression, we decided that should be our ultimate vision. 

We have a working prototype for conversational storytelling. Curating and sharing them out will be next.

When our project started, we imagined a tool for a specific animation, “Dragon’s Delusion,” created by our team members KongKee and Lawman Lok, and that is where the project name came from. With two artists and three technologists on the team, we taught each other about the animation process and the progress in AI. We saw the creative and tedious parts in making an animation, and identified where AI could be useful in speeding things up and making iteration cheaper. 

Thanks to the C/Change grant, we thought beyond ‘faster and cheaper’ and took on a humane lens in considering how to use the technology. Our team started by engaging with outreach programs for San Francisco’s underprivileged youth groups to experiment with AI tools. The game-character-ridden-AI-generated fictions the students created were fun, but the stories about themselves more revealing. We were compelled to build a platform for telling one’s life stories and sharing them.

Lessons

In 2023, we witnessed explosive growth in AI, and we certainly saw a plethora of competition for AI-aided creation and editing. When interviewing some documentary filmmakers, we heard loud and clear the urgent need to stylize protest footage to protect identity. Yet we had to leave that request and focus on our path. It was hard to say no, and we had to do that to many good ideas with good causes.

Future

History shall not be a text from a single source (often the winner). It will be supplemented by stories, and summarized into fables by the volume and precision AI can enable. Dragon’s Delusion may no longer be in service, but these stories live on to seed future fables.

* The opinions expressed in the project documentation are those of the participants and do not necessarily reflect the opinions of the Goethe-Institut San Francisco, Foreign Office of the Federal Republic of Germany, or Gray Area