Question

A retail bank is deploying an AI-powered virtual assistant that customers interact with through a voice-only phone interface. During a responsible AI review, the AI council notes that customers who are deaf or hard of hearing, as well as those with speech impairments, cannot use the service at all. Which Microsoft responsible AI principle is MOST directly at issue, and what should the council require before approval?

Accepted Answer

Inclusiveness — require alternative interaction channels (e.g., text chat or TTY support) so people of all abilities can access the service.

Answer

Transparency — require the assistant to disclose to callers that they are speaking with an AI system.

Answer

Reliability and safety — require additional load testing to ensure the voice service stays available during peak call volumes.

Answer

Privacy and security — require voice recordings to be encrypted and retained only for the minimum necessary period.