There was a screen inside the Meccano contraption. I think that when you moved the Meccano frame, it would rotate the axis and the picture inside the screen would move at the same time, thus displaying different angles of the picture. Not 3d but fake perspective.

A bit like the illusion with the Wii remote and the targets moving in sync from Jonnhy Lee.

Yes, or math, for example you can tell him to "add a+b where a=3.274914 and b=2.4847".

I doubt this precise numbers are in the dataset of chatGPT and yet it can find the answer.

According to this paper it seems to have gain the ability as the size of the model increased (page 21): https://arxiv.org/pdf/2005.14165.pdf

" small models do poorly on all of these tasks – even the 13 billion parameter model (the second largest after the 175 billion full GPT-3) can solve 2 digit addition and subtraction only half the time, and all other operations less than 10% of the time."

That's crazy.