I dont believe that is right. I think it was a university in Aus (Sydney?) that used to be quite heavily involved in this. From memory 100 volunteer subjects were given 50g in carbs from
a thing in isolation, and the effect on their Bg levels was measured at intervals over the next so many hours, and compared against pure glucose (GI 100). The 100 results were then averaged out to try to take account for the individual variance you mention
This is also a weakness of GI as a measure. Things are rarely if ever eaten completely in isolation. And also you don’t always eat 50g in carbs of everything.
There was an adjusted measure called Glycaemic Load, which tried to factor in likely portion size.
glycemicindex.com
EDIT: the website says ‘10 or more’ people, rather than my memory of 100