Jan 2022 - May. 2022 (Launched)
Product Designer
Pranav Chopada, Zach Dive, Jahnavi Jambholkar
Product design, visual design, product strategy, user research
The primary education system fails to keep up with the changing technologies of our time and lacks the engagement and personalization needed to spark creativity in children at a young age. Tales AI is a story generator that aims to make class readings more engaging and interactive by involving students in the story and adding fun to learning.
In the User Interface Design and Development course at UC Berkeley, I worked closely with my team to create the Intelligent Application by following a complete product design and development process.In addition to leading UX/UI design and product strategy, I also contributed to front-end development. This experience has inspired me to view design from a higher level perspective.
Showcased at Jacobs Institute for Design Innovation.
Launched, try our live demo.
Story books have long served as a way to expose children to limitless creativity and teach them the moral ethics of the world while building vocabulary and encouraging conversation and engagement. Unfortunately, these stories lack involvement (not personalized) and creativity that can be helpful in the growth and development of children.
Fewer children are reading books frequently for fun, according to a new report released Thursday by Scholastic, the children’s book publisher. In a 2014 survey of just over 1,000 children ages 6 to 17, only 31 percent said they read a book for fun almost daily, down from 37 percent four years ago.
Reference: https://www.nytimes.com/2015/01/08/us/study-finds-reading-to-children-of-all-ages-grooms-them-to-read-more-on-their-own.html
Story books have long served as a way to expose children to limitless creativity and teach them the moral ethics of the world while building vocabulary and encouraging conversation and engagement. Unfortunately, these stories lack involvement (not personalized) and creativity that can be helpful in the growth and development of children.
Participants: 2 Parent, 2 Children
Method: Facetime Call
We conducted our contextual inquiry virtually over Facetime calls in the evening as the participants were about to read a bedtime story to their child or the children were about to read a story before bed. During the calls, the participants were asked to share their experiences of reading stories and their thoughts on the current options available for children's literature. They were also asked to demonstrate their current methods of reading stories to their child, including the types of books they used and how they engaged their child during the reading process.
These photos are generated by Midjourney.
Our goal is to create and develop personalized user stories through minimal user effort. This is possible by leveraging intelligent systems (GPT-3.5, Dall-E-2) to create engaging, personalized stories for children to stimulate imagination and foster learning. The application will be inclusive with an option to create visual stories with a simple button click. We plan to incorporate voice-based story narration for people with special accommodations if time permits.
Generative AI for stories is gaining popularity. Companies like Jasper, Novel AI, and Sudo Write cater to different audiences and are continually evolving to serve a broader user base. But existing applications often fall short in harnessing the potential of technology to foster creativity and active participation for kids.
We filmed our video to reflect the process of how we imagine Tales.ai will be used. Our mission is to provide an innovative and interactive educational experience for students by leveraging generative AI to create personalized storybooks that engage and educate young minds. We want to empower students to unleash their imagination and become lifelong learners through the power of personalized storytelling.
In designing with GPT3.5 and DALL-E 2, we set up three prompts to help build a system for generating stories for children:
User Prompt: This is what the user enters. In our design, users are required to enter whatever they like as the "topic," set the story genre, and provide a character name. To make sure the story content fits better with the UI design, we limited the maximum number of words in the story and story title and incorporated them into the user prompt.
System Prompt: This is an initial prompt to set the behavior of the model. In this case, we set the goal as "educating students on the topic they enter through interesting stories in the selected genre."
Image Prompt: When we get the generated content, we slice it into different paragraphs and generate an image in the selected style for each paragraph.
To create engaging storybooks for children, I researched different image styles and story genres that are popular among children. I also tested different prompts for generating images and stories across various genres to determine the most effective ones.
We set up Dall-E and GPT-3.5 turbo api calls to generate the stories and images. We used MongoDB to store user and story information and AWS S3 to store all the images of the stories. For the frontend framework, we chose React to implement all the web app interface.
During the implementation process, I was was responsive for the front-end development of the story generate page, and the CSS code of the whole project to make sure the visual is consistant with the user interface design. Although I was new to front-end development, I have gained valuable knowledge in implementing UI design. These lessons have helped me to better understand the engineering work.
01
02
03
04
05
06
07
08
09
