Hi.
One of lifes main inputs is through the senses.
If Ziggy/Hal is given a voice analyzer to tell differences
in the tones of voices then the bot could tell If more than one person is speaking. Without an analyzer Hal cannot tell if there is more than one person speaking, an analyzer can lead to Hal waiting his turn in a triple/+ conversation to say something to either person he is listening too.
If a really good stereoscopic microphone is implimented with some really good speech recognition software then you could set the bots ears in front of the T.V set so he can learn from Television programs.
This can be enhanced by giving the bot a good set of stereoscopic eyes with good visual recognition software.
but in order to make it work these things must be high in quality.
Best of wishes and brand new discoveries.
Jerry.