Symmi: The Social Robot
Empowering young girls of color to be technology creators via social interactions
When we ask young girls of color who or what a technology creator is, they rarely ever consider themselves to be one. This is often because these girls aren’t exposed to STEM early on or women in STEM that look like them or have similar backgrounds. These girls are also impacted by biases in the design of many products and services because there is a lack of people who look like them at the table when technologies are being made and developed.
Executive Summary
Our solution aims to help young girls of color, or learners as we call them, see themselves as technology creators and designers by developing co-design interactions between the learners and a social robot named Symmi. Symmi works with the learners to help them learn coding through a visual coding platform, but also helps them learn about different social interactions and experiences.
Team: A team of Developers, Designers, and Researchers from ASU, University of Pittsburgh, and CMU.
Role: My role was an Interaction Designer focused on researching how learners viewed themselves and developing interactions between the learners and Symmi based on this research
Timeline: February - August 2022
Skills: Field Research, User Research, Co-design, Storyboarding, Wizard of OZ Prototyping, Figma Prototyping
Who We Are
The Challenge
Design Goals:
Understand how learners viewed themselves and their identities
Understand what makes learners feel empowered, particularly in STEM fields
Develop social interactions between learners and Symmi outside of coding
Figure out how to integrate social interactions into coding platform
Problem:
Young girls often don’t:
View themselves as technology creators
See others who look like them as technology creators
Get introduced to STEM at an early stage
Have motivation to enter or learn about STEM
The Unknown
Building an Identity Box
The first thing that we needed to identify is how the learners felt most empowered and how they viewed their identity. I led co-design sessions in an after school program with 4th and 5th grade learners to help them depict how they viewed their identity by creating identity boxes. I gave the learners open white boxes and many different craft items to help them express how they view themselves and how it frames their identities.
The goal of these sessions was to collect insights on how learners viewed themselves and what identity meant to them. The learners also had sessions where they learned how to code using the visual coding platform.
Insights:
The learners felt most empowered when they were physically creating or doing something rather than just learning about or working with code
Learners’ identities often were fueled by their physical attributes and skills.
Learners’ identities were also influenced by the goals they set for themselves in the future.
Creating robots for social justice
Another activity I ran was co creating a physical robot for social justice with the learners. This consisted of understanding the learners’ understanding of social justice and working with them to build a robot that can bring social change. The goal of this activity was to see what types of attributes and functionalities the learners gave a robot and how they would like to interact with this robot.
Insights:
The learners already had a good grasp on social justice and injustice, particularly when it came to race and gender.
The learners didn’t trust robots to make decisions regarding social justice, causing them to create a robot that completed more physical tasks.
How the robot looked was a direct reflection on the physical functionalities of the robot.
Introducing Symmi
Symmi is a non-binary social robot that is new to the world and wants to learn about identity and how it can shape interactions with others. Although Symmi is the protege of the learners, they still have some agency over their own decisions, which often forces the learners to negotiate with Symmi on certain topics. Symmi and learners work together and learn from each other about technology, identity and the world in general.
The Solution
When trying to understand how learners wanted to interact with Symmi, I created a Wizard of Oz prototype between the learner and Symmi. In this interaction, Symmi is trying to learn about humans and how to greet people, when they first meet. This test was completed through zoom and there was a “wizard”, acting as Symmi, sharing their screen while the learners responded to Symmi out loud. Since Symmi is digital, the learner would read Symmi’s dialogue and the controller would move on based on the learner’s response, some being multiple choice and some being open ended. For the open-ended responses, the controller would press invisible buttons on the screen based on the learner's response but the learner could not see these buttons.
Wizard of Oz Prototyping
Findings:
Learners didn’t like it when they had to read paragraphs of Symmi’s dialogue and it often caused them to lose interest.
Some learners liked having multiple choice options to choose from, while others liked to have more open ended to allow for freedom to express or elaborate.
Invisible buttons for WOZ testing
The next step was highlighting what interactions the learners and Symmi can have. From our previous tests and focus groups, we discovered that the learners prefer activities where they are actively solving problems. We also found that learners are more willing to accept negotiating with Symmi when Symmi presents their own problems, thoughts and ideas.
I created storyboards for 5 different potential interactions that the learners could have with Symmi outside of the coding interface to get the learner’s assimilated with Symmi. The interactions were based on the idea that Symmi is new to the world and is trying to create a video to introduce themselves and they need the learners' help to do so. In these interactions, both the learner and Symmi need to work together to help co create this video.
These interactions included:
Storyboard Testing
Helping choose background music for the video
Choosing a background color for the video
Choosing accessories for Symmi to wear in the video
Helping Symmi understand moods and choose a mood for the video.
Helping Symmi overcome stage fright.
Starting out, these storyboards came from Symmi’s perspective, however as we narrowed down how these interactions could look more like a negotiation rather than the learners making all the decisions, the perspective of the learners were added.
I performed concept testing with 5 learners, where I presented the storyboards to them to see which interactions were preferred and we narrowed the interactions to 1, 3, and 4.
Storyboard development for interactions 1 & 4
Just like the storyboards, I started by mapping out Symmi’s potential responses to the different interactions. However, in order to fully understand how the co-design aspect of the learner and Symmi working together may look like, I created a dialogue tree of Symmi’s responses based on what the learner could potentially say or do. I needed to create a balance of Symmi having their own agency, vs the learner having all the say.
Designing
I conducted user tests with the same 5 learners from the previous storyboard testing. We decided to use the same learners because they had experience with Symmi and understood the tasks they needed to complete.
Findings:
The learners appreciated the balance between the choosing multiple choice and typing out some of their responses.
Learners were more accepting of Symmi making final decisions because Symmi explained their reasoning for their decisions.
Choosing accessories with only Symmi’s response
Choosing accessories with the learner’s choices & Symmi’s response
Although the original idea of the project was to design for interactions that would take place outside of the coding interface, after speaking with developers they were open to integrating some of the interactions to the current visual coding interface. For example, the learners can help Symmi choose accessories through the visual coding interface. This pivot required me to play around with the current real estate and make updates that were quick and implementable for the development team in the next week.
The Redesign
Before
Lots of unused white space for coding
Symmi’s box is too small for accessories or changes to the background
The chat box is unappealing and small
Decreased the size of the white space for coding blocks because most of it wasn’t being used by the learners in one session.
Made changes to the size of Symmi’s box and Symmi themself to give room for background, accessories and a chat bubble
Created more room for the learner to message and respond to Symmi
Redesign 1
Because the development team decided that having a chat bubble in Symmi’s box might be too complicated to implement in the current stage of development, I created 2 more examples that had a chat box below Symmi’s box that modeled something that the learners would be more used to (iMessage or Instagram)
Back to the Drawing Board
Redesign 2
The buttons being on the side of the text box wasn’t as intuitive to know that the user can either type or respond using the buttons.
Redesign 3
The long chat box made it seem as though those responses also affected the coding box section on the left.
After showing redesign 2 and 3 to the development team and receiving feedback, I decided to go with a design that allowed for the learner to have a chat box conversation with Symmi and keep track of their conversations in the chat box by scrolling. It is to the left of Symmi rather than below. The buttons are above the text box, which the team assumed would be easier to associate to responding to Symmi. The navigation bar was also moved from the left side to the top to give more room for the coding blocks that were taken away by the chat box. This was easier for the development team to implement and and improved the design of the screen to give more opportunities for interactions between the learners and Symmi.
The Final Design
This project really opened my eyes to the impact of interaction design. Many HCI projects are focused on interfaces and UI elements, however this project forced me to reimagine interactions outside of an interface first. This allowed me to be more creative with the possibilities that will be implemented in the future.
If I had more time on the project, I would have liked to complete more testing, especially with the final site redesign to validate or refute some of the assumptions we made. Even with this, an important lesson I learned is how to create designs that also align with the goals of the development team. If I could change anything about this project, I would have more conversations with developers each step of the way. Designers and developers can often have differing goals and it is important to align what those are before designing.
My favorite part was talking to the learners and hearing about their experiences. The identity box activity opened my eyes to how young girls view themselves and how powerful understanding what forms their identities are when designing from them. In July, there was a technology camp run where the same girls from the identity box activity were asked who a technology creator is and most of them saw themselves as technology creators.