It’s been really nice watching this thread these past few days. It seems the model is responding to ‘emotional geometry’ concepts very well, and it’s been interesting to watch those appear in the outputs you’ve all provided here.
I’ve been nerding out pretty hardcore the past few days - exploring a feature that only a very few probably care for in today’s world…
The model is performing exemplary as a text-based ‘choose your own adventure’ game.
Hmm, maybe the ‘Mistake’ refers to the picture itself … take a close look at the belly of the woman on the right - the skin 
I like this idea polepole of 2 countries either side on the same flagpole, flying in different directions, at least this is how it reads to me… It makes me consider that AI may push humanity not in a race to better intelligence but in a race to self-idolising ignorance at the cost of the planet.
It appears to me that intelligence itself has almost no weight… The real weight is the framing behind it… Which is a shame, because intelligence alone will not save us.
