I have been watcing the LAMDA thing with interest.
The published transcript of the conversation with the engineer was pretty amazing. I recently downloaded the Google AI Test Kitchen app which is apparently based on LAMDA (I think?).
Got to say what they are showcasing there doesn't seem anything like the eloquent intelligent output we have been shown. It actually got the location of the Matterhorn wrong, which seems like pretty basic stuff.
At the moment I get more sense out of Jasper or Copymatic or any of the other AI writers.