Voice User Interface Design

The goal of this project was to empower and support older adults by leveraging the accessible benefits of voice technology. Throughout each stage of the process, we followed inclusive design principles.

As the design lead, I navigated my team through the complexity of designing for voice by creating our design process. This involved establishing design standards and principles, and developing ideation and design exercises that allowed us to translate our research into a successful solution. 

CONTRIBUTIONS

• Design lead

• Generative research

• Usability testing

• Project management

Summary

Given the recent advancements in voice technology and the accessible nature of voice interactions, we wanted to explore how a hands-free solution like Amazon’s Alexa could help meet the needs of semi-homebound, older adults (people aged 60 and older who are not able to leave home without difficulty or another’s assistance). Individuals who experience changes in physical mobility, reduced social connectivity, and memory, speech, and hearing decline have more difficulties leaving their home. While voice technology has the potential to benefit semi-homebound, considering the needs of “extreme users” ensured that our solution was inclusively designed.

Our research on Maslow’s Hierarchy of Needs indicated that self-esteem is critical to the wellbeing of semi-homebound, older adults. The top three factors that appear to affect self-esteem include independence, productivity, and feeling challenged. To help our target population meet these needs, we developed a conversational model that engages people in their goals and interests. Unlike existing voice user interfaces (VUI), the VUI for the “BOX” conversational model was designed so it would be accessible to people with memory, speech, and hearing decline.

Our final deliverables include a conversational model, best practices for inclusive VUI design, and a video-sketch prototype.

This project was sponsored by the Amazon Alexa team.

Design Problem

The number of older persons (60+) in the world is expected to double in size by 2050, to approximately 2.1 billion people. While there are technologies that can assist and support the aging process, many geriatric technologies focus on physiological needs (such as prescription reminders), which is only one subset of possible needs for an individual as they age. Research points to a discrepancy between older adults and their caregivers in the actual vs. perceived needs of semi-homebound older adults, and that existing technology does not adequately address this gap.

In addition, recent technological advancements in artificial intelligence and natural language processing have created opportunities for more complex, and richer voice interactions between humans and computers. Although still in its infancy, VUIs show promise as being a more natural way to interact with computers. However, today's voice technology is optimized for the "average user", and although older adults may significantly benefit from VUIs, the technology is unable to adapt.

Findings

Through our research, we had three key findings:

1. Navigating graphical user interfaces (GUIs) is challenging for older adults due to inaccessible design, and the fine motor movements and coordination required to use computer keyboards, touch-screens, and mice.

2. Esteem needs are a key unmet need for older adults due to physical and mental impairments that result in limited independence.  

3. Access to technology is just as important for older adults as it is for everyone else. It is particularly important to older adults for social connectivity and autonomy. 

Solution

Our findings validated our decision to build a solution using a VUI rather than a GUI. In addition, our research indicated that we could make the greatest impact on our target user group, by designing an experience that focuses on self-esteem. Many of the older adults we worked with experienced lower esteem due to the lack of control over their daily agenda, which was often a result of being homebound.

Our final solution is a conversational model that:

1. Understands the limitations and needs of its user.

2. Connects to and accesses many information sources on the internet.

3. Is optimized to help users pursue their goals and interests via a VUI. 

This model was designed using the best practices we defined for inclusive VUI design. You can view the best practices booklet here.  

The video below demos the conversational model we designed.

Key Challenges

1. Establishing the design process. Because designing for systems such as Amazon Alexa is still so new, I developed a lot of my own methods from limited resources and have been "play-testing" the process. I used a combination of tools from the dramatic arts (improv, table reads, acting, character profiles) and interaction design (flow charts, user stories, design sprints, participatory design) to take my team from research into prototyping. Establishing our design process allowed me to more intimately understand the role design plays in creating VUIs, and exposed me to a breadth of ideas in the field.

2. Designing with robust standards. Because VUI is such a new space, the best practices are still developing and are not entirely inclusive (especially when it comes to accessibility). To address these limitations, my team conducted our own best practices research with our target user group, and have cross-referenced best practices from Amazon and Google to ensure strong design standards for our solution. In addition, besides formal documentation, one of the neat things that I've realized about voice design is that the gold standard for voice interactions already exist in nature. So simply comparing your VUI with how humans actually communicate is a pretty fail-safe way to ensure you are moving in the right direction. Brings a whole new meaning to human-centered design!

3. Prototyping a solution that is not confined by existing technical constraints. To overcome this, we used the Wizard of Oz prototyping technique to design solutions that are not limited by current Alexa technology.

4. Gaining knowledge in a number of subject matters. This project required our team to have a strong understanding of a number of subject matters in order to develop a successful solution. This includes older adult health and behavior, voice technology, internet of things (IoT), and the psychology behind esteem. We conducted multiple literature reviews and expert interviews to help support this process.