Case study: Define and prioritize

Preview: This Pre-GA product is subject to the "Pre-GA Offerings Terms" in the Service Specific Terms, offered "as is" with limited support. See the launch stage descriptions for more details."

The following describes how we applied the definition and prioritization phase of the Natively Adaptive Interfaces (NAI) approach to a specific use case: making online videos accessible to blind and low-vision (B/LV) users.

1. Establish a viewpoint.

   "Blind users need to access information available from online videos because how-to video guides are an increasingly important medium for learning."

2. Formulate multiple point of view (POV) statements.

   • "Blind users need to access video entertainment independently because it can help improve mental health and social connection."

   • "People with low vision need customizable descriptions of visual details because standard AD might be too much or too little information."

   • "Blind users need online videos to be accessible because videos as a medium has become a societal resource for learning and entertainment that all people should benefit from."

3. Evaluate and select the best POV.

   Best POV: "Blind users need online videos to be accessible because videos as a medium has become a societal resource for learning and entertainment that all people should benefit from."

4. Brainstorm features.

   Identified the following features to create equivalent input and output experiences.

   • Input features:

     • Adaptable preferences: users can provide feedback to the agent, allowing it to personalize the interface, prompts, and interactions based on individual preferences and context.
     • Information seeking (Q&A): users can ask questions about the video content at any point, whether directly related to the visuals or exploring broader or tangential topics, such as where to buy a featured product or the definition of an unfamiliar word.
     • Playback control: beyond classic voice-controlled playback, users can navigate the video using natural language, like jump to the part where they discussed the recipe.
     • Agentive screen reader: instead of reading what's under focus or what the user navigates to, it would anticipate needs, understand context, and take initiative to read what it believes the user will most likely want to engage with.

   • Output features:

     • Personalized guidance: the agent is able to offer adaptive support based on user preferences and interaction history.
     • Adaptable audio descriptions: the agent provides adaptive audio descriptions that allow users to select a delivery style, for example, detailed or fast-paced, that suits their preferences and the content.
     • Information retrieval: the agent retrieves information from the video or external sources, indicating the origin of the information. The agent provides summaries and answers to questions about the video, including broader topics.

5. Engage with users and gather insights.

   Conducted 20 co-design sessions, 16 with blind users and 4 with low vision users. Feedback obtained from each session was immediately triaged, and improvements were rolled into the next version of the prototype. The goal was to rapidly iterate through prototype versions on a weekly cadence.

6. Prioritize features.

   After completing all co-design sessions, created the following list using a priority scale of Highest, High, Medium, and Low:

   • Input features:

     • Adaptable preferences: High
     • Information seeking (Q&A): Highest
     • Playback control: Highest
     • Agentive screen reader: Low

   • Output features:

     • Personalized guidance: High
     • Adaptable audio descriptions: Highest
     • Information retrieval: Highest

What's next

• Key terms