Product design question

Design a phone for the blind.

Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.

What this question tests

Whether you can design for an accessibility-first use case where the interaction model itself, not just visuals, must change.

Clarify scope: a phone designed primarily for blind and low-vision users, not a sighted phone with an accessibility mode bolted on.
Identify core needs: non-visual navigation, reliable text-to-speech, and tactile or audio feedback for every interaction.
Prioritize voice and gesture as the primary input method, since a screen-based touch grid is hard to navigate without sight.
Propose a concrete interaction model: consistent spatial gestures paired with audio confirmation for every action, plus a dedicated hardware button for help.
Address a key risk: reliability of voice recognition and text-to-speech in noisy environments, since failures here break the whole experience.
Define success metrics: task completion rate for core actions like calling or messaging, with a guardrail on voice recognition error rate.

Designs the interaction model itself around non-visual input, not just a screen reader layered on a sighted UI.
Prioritizes reliability of voice and audio feedback as the top constraint, since any failure there blocks the whole phone.
Proposes a specific, consistent gesture and audio-confirmation system rather than a vague more accessible phone.
Names a guardrail, voice recognition error rate, since accuracy failures directly break trust for this user base.

Designing a standard smartphone with accessibility features added on top, instead of rethinking the core interaction model.
Ignoring noisy real-world environments where voice-based interaction is most likely to fail.

How would you test this with actual blind users rather than assuming their needs?
What would you do for the subset of tasks that are hardest to do without sight, like reading a physical document?