Product design question

Design a calculator 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

Accessibility product design requiring a genuinely different interaction model, not just a voiceover layer.

Clarify the core barrier: a visual calculator relies on seeing input and output together, which a screen reader alone handles awkwardly for multi-step math.
Segment users: blind users comfortable with braille displays versus those relying purely on audio.
Identify the pain point: verifying a long calculation by ear is hard, since audio is sequential and users can't glance back at prior steps.
Prioritize an audio-first design that reads back each entered digit and operator immediately, plus a repeat-last-calculation command.
Design the solution: voice input for numbers and operations, audio confirmation per keystroke, and braille display support where available.
Define success: calculation error rate and completion time compared to a sighted user's calculator use.

Identifies the real accessibility gap, verifying multi-step sequential audio output, not just adding voiceover.
Segments by assistive technology used, showing nuanced understanding of the blind user population.
Proposes concrete interaction details, per-keystroke confirmation and a repeat command, instead of vague accessibility language.
Uses error rate and completion time as objective, comparable success metrics.

Assuming a simple screen-reader layer over an existing visual calculator is sufficient.
Treating all blind users as having identical needs and tools.
Not defining an objective way to measure usability improvement.

How would you test this with actual blind users rather than simulating blindness?
How would you handle scientific or multi-step calculations differently from simple arithmetic?
What tradeoffs exist between speed and clarity in the audio feedback design?