Sme issues are not as simple as some would like to see them. Following is for the non-religious (i.e. for those w/o preconcieved ideas of digital photography).
1. Of course, converting from 8bit to 16bit does not add anything to the image data except for higher storage requirement,
2. if several steps follow, which incure interpolations, the conversion may prevent certain losses. I would not venture to quantify the possible losses; I venture to say, that anyone quantifying or disquantifying the losses is an amateur not deserving an answer.
However, the possibility of cumulated loss (or gain due to the randomness) should not be ignored.
Example: I am creating a panorama. I am starting out with raw image frames. Normally, I am converting them in 16bit ProPhoto; assumed that I had not done so, the question is if it offers any advantage to do so afterwards.
So, what steps are involved in the coming process?
a. Warping the frames (called "stitching" by those, who don't understand it). Interpolation pure.
b. Blending the resulting frames. Interpolation pure.
c. Post processing the pano: selective contrast and brightness adjustments and sharpening. Sometimes trying to repare what has been blown in the previous steps. Anyway, lots of interpolation.
Now, one can say justifiedly, that thios is not the normal process for most photogs. However, other photogs are using their favourite HDR, sharpening or wharever pluf-ins. Again, lots of interpolation.
Now, substitute interpolation with loss of tonality/detail.