Unfortunately while Posits (and Unums) are more accurate for some computations, they are still designed around using small numbers of bits to represent numbers,

Unfortunately while Posits (and Unums) are more accurate for some computations, they are still designed around using small numbers of bits to represent numbers, so they have error in computations just like floating point. The argument for posits are mostly that they produce less error on the inputs you care about, and more on the ones you don't, but it really depends on your computation; you really have to know what you're doing to use them safely.