I appreciate Lanier’s reminder that AI is ultimately a collective artifact, not an independent mind. Calling it "large language social collaboration" might be clunky, but the point is fair: what we label "AI" is just our accumulated digital patterns, recombined.
I do wonder, though. At what point does the remixing and generalization become more than just a mirror of us? Even if it is still trained on human work, its behavior can surprise us in ways that feel alien.
Curious if others here agree: is the term "AI" just marketing hype, or is it still useful shorthand for systems whose emergent properties we do not fully control?
My objection to calling it collaboration is that the contributors were not given a choice of whether or not to participate.
To collaborate is to voluntarily join with others
and contribute willingly.
The current crop of LLM models were created from large-scale appropriation of the works of others without consent.
This is a poignant observation. What if there was real tender love and care put into gathering these training materials? (Y'know, truly free and open source, with a deep understanding of what it means to be human.) If you are working on this, we should talk.
