In the old days, the apprentice would pay a fee to the master, and "In theory no wage had to be paid to an apprentice since the technical training was provided in return for the labour given. However, it was usual to pay small sums to apprentices, sometimes with which to buy, or instead of, new clothes." Labor is a complex arena, and I am no expert, so I appreciate your insights. I find that the binary categorization of such a variety of interactions as "paid/unpaid" with "unpaid=bad" is an overly simplistic way of rendering judgment. NACE, the source of the data behind that chart, repudiated the conclusions made in the article, describing it as "misapplication" of their survey results.
In the old days the apprentice also got room and board. "Wage" meant something else. I grabbed that graph off the Internet flippantly. However, considering the data it was drawn from, those are interns that were also earning college credit. I have friends that have burned through six of those peeps. This is the sort of economy that allows a couple of Youtube producers with truly marginal earnings access to a student pool that pays over $20k a year in tuition... for free. As an education it's definitely beneficial and as interns they're damn handy but there is a value proposition in there in place of wages. The unpaid internships highlighted in the article would not be covered under this arrangement.