But these commons are now being overgrazed by rapacious tech companies that seek to feed all of the human wisdom, expertise, humor, anecdotes and advice they find in these places into their for-profit A.I. systems.
I think that the concept of tragedy of the commons is being misused here. When feeding data into those models, there’s no common resource being used, as the data doesn’t cease to exist once you feed it to your L"L"M. Instead what’s happening is that they’re further breaking what was already broken - the legal concepts of IP and copyright.
Where the concept could apply is the usage of the output of those models, with the common resource being the overall quality, reliability, and usefulness of the internet, for the sake of petty benefits (such as advertisement/spamming/marketing). However this degradation predates the large “language” models (and the internet itself), and it isn’t a result of the technology itself.