Because it allows people to communicate when they're not in close physical proximity. Would you rather go out to dinner with friends and just speak to each other or sit there and type your conversation out in a WhatsApp group chat?

It's a convenience/necessity thing, pure and simple.

Theres benefits to be had when interacting with REAL people in person.

Zero benefit interacting with voice with an AI. Pure and simple.

Nobody cares about an agent when they are the principal - this is not remotely the same as interfacing with a human that is valued much higher.