Voice is also conversational if it’s two-way.
I increasingly don’t want AI to reply in words, only in actions.
It’ll likely be a constantly changing diffusion model. Like a dream, basically