OK, now it makes sense. But that would be a huge problem to overcome. Finding emissivity/absorptivity (or inverse, reflectance) values for human skin took very little effort:
http://repository.upenn.edu/cgi/viewcontent.cgi?article=1616&context=cis_reportshttps://www.imt.liu.se/edu/courses/TBMT36/pdf/Text4.pdf Both studies clearly show that absorptivity is actually extremely good but specifically tuned for infrared spectra and not visible spectra, which probably makes intuitive sense. Arcs are so hot that they are just going to have an emissivity spectrum of a good white body radiator so we only need to know or at least estimate temperature.
Still, the measurement data tells us something if you glance through the above reports. Human skin for whatever reason is pretty close to a good black body. It looks like absorption in the visible spectrum is about half of what it is in the near infrared but that's not really all that significant. I'll bet that the result would be nearly the same for a face shield irrespective of whether color spectrum is taken into account or not. Over the range of wavelengths of interest, copper calorimeters are not very discriminating either.
My very unscientific look here seems to suggest that even if we can more closely mimic radiant heat absorption by human skin, it won't change the result significantly.