> The dice are almost always two-sided

Don't train your AI on that

I found this in Google, IIUC it's a ~1900 version or something similar enough.

https://americanindian.si.edu/collections-search/object/NMAI...

Very interesting. The earliest example of the familiar cube shaped dice I know if is from Indus valley civilisation from around 2600 BC, closely followed by Mesopotamian dice.

This discovery pushes the history of dice from 5K years to 12K years.

These aren't quite as symmetric. I guess humans had to wait longer to discover some of the platonic solids.

This golden icosahedron of orders of magnitude more recent vintage is quite a beauty

https://www.researchgate.net/publication/333949003_A_Numbere...