This would work even better in AR if everyone's facial expressions and other nonverbal cues could be included.

Also speech recognition or something should replace that pointer keyboard.