It's already here: https://youtu.be/ixgFtjfO_7Q&t=593
People are combining GPT3 with another AI that translates between text and verbal speech through an avatar. You talk to the avatar, it converts speech to text and sends it to GPT3, GPT3 responds, the avatar says that response out loud.
This clip I linked to shows two separate instances of this AI configuration, each talking to the other.
If you're asking about interaction in the sense of what Kitt did in Knight Rider, or what Jarvis did in Iron Man, then we're about 6 months away. Embedded systems integrating this tech with real world action is a small gap to close, and will be the next money maker for entrepreneurs.
(post is archived)