Team /
Nelly Shih
Kay Zeng
Yihan Yu
Role /
Researcher
VUI Designer
Usability Testing
Timeline /
9 weeks
ALFRED
Improving the coordination of schedules and spaces in a household
INTRODUCTION
Alfred is a home voice assistant that brings everyone together. In response to economic and public health concerns brought on by a pandemic, the increased amount of time spent at home, and the transition to remote working and learning for many professionals and students, we developed a VUI utilizing familiar, existing technology to improve communication between household members and facilitate more productive working/schooling conditions within the home.
BACKGROUND
Over the last two decades, the number of young adults cohabiting with people other than their partner or spouse has steadily increased due to student debt burden, rising housing costs, and trends in delayed marriage. With the COVID-19 pandemic, health and economic impacts drove even more people to reconsider their living situations. A July 2020 Pew Research Center report found that 52% of young adults have returned back to their parents’ home. Parents and their adult children have had to balance a professional remote work or school schedule within the confines of their new housing arrangements, in addition to navigating familial interpersonal dynamics and re-establishing co-living boundaries. This informed our focus on increasing productive communication between members of a household to facilitate remote working/learning and improve cohabitation.
RESEARCH
Utilizing mixed methods, we focused on adults who currently live or had recently lived in a parent/children household. The family dynamic presents unique relational challenges that may make it a more extreme case compared to a household of non-familial roommates. We surveyed 39 young adults who had moved in with parent(s) within the past 8 months.
Key Takeaways
-
Common issues to remote working/studying included a lack of quiet environment, creating makeshift workspaces within shared spaces, and frequent interruptions from family members.
-
Most indicated saving money as a reason for moving in with their parents. But they also saw it as an opportunity to spend more time with family.
We also conducted individual interviews over Zoom with 6 young adults and 2 parents living in multi-generational households. Remote interviews provided a unique opportunity to observe participants within the environmental conditions we were asking them about. From our interviews, we organized findings into these recurring themes:
Disconnect in recognizing needs of people in “work mode”
- Young adults experience frequent micro disruptions (being asked for tech support, called to join a meal, or engaged in spontaneous conversation) by family members while working or attending class.
- One participant was interrupted by family members coming in and out of the room multiple times during the interview despite her clearly being on a video call.
Communication patterns are household-specific
- Families have informally organized different ways of communicating their schedules with one another but all have experienced some form of miscommunication or conflict over schedules.
Managing family time vs desire to be productive
- Most young adults appreciated the opportunity to spend more time with family but they struggled to maintain their previous levels of productivity prior to moving due to increased family distractions.
IDEATION
With these insights in mind, we brainstormed and sketched ideas, mapping them along the following axis:
Our design had to support multiple users as it was centered around the family unit at home. It also had to be accessible to people with low to medium tech literacy. We integrated a few of our ideas under the concept of Alfred, a voice user interface (VUI) that would serve as a family scheduler.
Adding to the functionality of existing smart home devices like Google Home, we developed a new feature that expands connectivity between shared devices and family
members. Currently, up to six accounts can be added on a Google Home device but family members are only able to access their own accounts with Google’s voice recognition. Managing a shared calendar requires maneuvering through different products and digital devices, which can be challenging for those with low tech literacy.
Alfred provides a collective view of each family member’s schedules based on the needs members want to communicate with others. With Alfred, family members can:
-
Set up a shared calendar to add individual events, such as work meetings or class schedules
-
“Reserve” a shared space in the home
-
Check one another’s availability
-
Create family events and send invitations
-
Set notifications and reminders that will alert everyone in the network.
PROTOTYPING
We designed a mock up of the conversation flow and used Voiceflow to build the interactions for accomplishing three core tasks with Alfred:
-
First-time setup
-
For calendar events, adding the ability to “reserve” a communal space in the home (ex. living room, office, dining room), choose an event mode (ex. Available, Limited Availability, or Do Not Disturb), and set a reminder that alerts everyone in the network
-
Checking family member’s availability and scheduling a group event
While designing Alfred's conversation model, we incorporated semi-formal English speaking conventions to simulate natural dialogue and referenced Google Home’s existing flows to maintain consistency. Alfred can recognize intents based on utterances we have compiled and provides clear prompts to maintain low cognitive load for task completion. The model includes corrective options for handling unexpected utterances, as well as confirmation messaging following complex conversation flows to allow users to recognize and recover from errors.
INSIGHTS
We conducted remote Wizard of Oz tests of our VUI prototype with 3 young adults currently living in multi-person households. A scenario was designed to test 3 task flows:
Task 1: Add family members and save a list of “shared spaces” in the home during set up
Task 2: Add a work event to the calendar as a notification. Set a mode for the event and reserve a shared space.
Task 3: Schedule a movie night with your family by checking everyone’s availability.
We observed their reactions for signs of confusion or frustration and took note of alternative utterances to understand their mental mapping of anticipated conversation flow. We also conducted post-task debriefs to assess the ease of interaction and task completion, and whether they understood the features presented by the VUI.
Participants’ overall experiences with the prototype were on par with their expectations of interacting with VUI systems in terms of conversation flow and logic. They indicated recognizing the limitations of voice activated technology and holding it to a lower standard of efficiency and accuracy in which may have affected their assessment of the experience.
“ Sometimes [Google or Siri] are not able to give me the answer I’m seeking… Sometimes I get more annoyed, and then sometimes it’s like, “Eh, they’re not advanced enough,” so I’ll let it go.” -P3
The new features provided by Alfred were clearly understood by all participants. All participants were able to complete all tasks without any guidance but most wanted more feedback from the system in terms of providing confirmation after long conversation flows or indicating additional options, like for a family member’s availability. Conversation flows were updated to address these areas in the interaction that created confusion.
NEXT STEPS
Additional testing and feedback of the prototype is needed with particular focus on parents living with their adult children and people with limited experience with VUI systems. Our scenario and tasks were also constrained due to the limited sets of predicted responses we could create. Using an AI-driven, large database of user utterances would allow for a greater and more natural range of scenario testing.