Right, and I suppose if you still tried to charge for use of references to source data, it would then be a weird slippery slope of weighting for which source data the AI was trained on first. How would you say, bill for references to a circuit board if it was trained on things like dictionaries that include “circuit board” as well as of course, more direct references to circuit boards in tech.

Guess it could be some weird percentage, but I don’t think I would welcome that reality