a phd is a little bit like "a 3 year marriage in the 1920es" . it means that it is 3 years of very close collaborations, so it is obviously important that the student and the advisor (and the team) get along well, have compatible ways of working, are interested in the same things ... and that the phd student has the necessary prerequisites, is sufficiently autonomous, has the right quality threasholds, etc. but, it is also a very asymmetric relationship: if things do not go well, the phd advisor in principle doesn't risk anything, whereas the phd student risks not graduating (that's why we compare it to the assymetry of marriages in the 1920's, where if a marriage failed the husband would still be the bread winner, whereas the wife would be "damaged goods" ). so, it is even more important for the phd student to ensure that the relationship with the advisor is viable, and that there's mutual trust.
so, we see an internship as a bit of a "trial period" -- not a matter of "getting extra work out of you" but for us (both the student and the team) mutually to get to know each other and figure out if there's a common project, trust and understanding, before embarking on 3 years together....and, if there's not, then stopping an internship after 3-6 months is perfectly fine, whereas dropping out of a phd program after 3-6 months is a potential disaster (in particular, for the student) ...