Question about microcontroller 8-bit, 32-bit?

The computer defines an 8-bit binary number as a byte, 4-bit is half a byte, then 16-bit is a word;

The definition of byte is unchanged nowadays, while there is some confusion about the definition of word. The data width of 32-bit, traditionally called 4 bytes, or two words, some articles on the cpu data width called word (32-bit or more), then 16-bit is called half a word (not a byte);

As for the address allocation problem, for compatibility issues, the current still assigns the address to the byte, so 16-bit width of the data in the storage needs two addresses, 32-bit will need 4

If future CPUs, such as 64-bit and 128-bit wide instructions, are only compatible with 32-bit, then 64-bit data will only take up 2 addresses;

So your understanding is correct;

This is a good idea.