The first are often in humanities and social sciences. They schlep stuff, track stuff down, organize conferences and make copies. Been there, done that. They're office employees, for the most part, assisting in another's research.

My "research assistantship" was exactly that. ILL, fetch things from URL (the library), precis articles, help organize a conference, do a lot of filing and deal with administrative crap.

I'll hedge on the "humanities and social sciences" bit, though. I knew grads in a psychology dept. (psycho-/neuroling) who cowrote papers and often got first billing (when alpha order wasn't required), who did side research on the data or who basically proposed the experiments to be run and the analysis as part of a grant (university or other) that was allied to the lab group's work but was their own baby. They were "employees" for funding, but even when "laboring" for their "boss" they still got academia cred (if not a academic credit--a grade and course credit) towards their graduation and post-doc. My wife was in that group--she didn't like the parts where she worked for her PI, but got publication credit; she liked the parts where she did her own thing, subject to approval by the PI (which she didn't like). Yet for 3 years she got a string of publications, conference papers, attended conferences for free and was even asked to give invited papers. Because she was a "laborer" and only $ counts. But all the other stuff was invaluable and returned a hefty investment. Wages weren't great, but higher wages = fewer researchers = less output. (Or fewer grants, and while a school might keep its contingent of student researchers there'll just be fewer grants so still fewer researchers and less output. And fewer researchers means fewer grad students, esp. at lower ranked schools with heftier socio-economic background mixes making their academic chops in preparation for the job market. So that the higher-ranked schools' students get more $. Esp. in this grant market where it's not a zero-sum game, like usual, but a negative-sum gum.)

The second category of researchers are often in science/engineering programs, where what they get out of the "labor" are publishable papers, poster presentations, and training on how to actually do the research, use the equipment, and learn the methods that they'll need in their careers but can't be taught in staid lab classes or lectures. Their work often helps their professors, but without that work they'd be hard put to be competitive for postdocs or often just not able to conduct their own PhD research, which often derives from their PI's work.

If they unionize and get representation for the latter, they should also sign up for a course that charges tuition/fees for the benefit. They help faculty but often get more in return than they give.