While "theft" is probably not a good descriptor of how an image synthesis engine is built, I think the accusation of "plagiarism" is a fair one. The machine is trained on the work of a vast number of artists sourced from multiple databases. We don't know who's in, but more importantly, we don't know who's out. I think it's reasonable to care about the loss of pedigree on the data.
... but it's worth noting that plagiarism tends to be more an academic concern than a corporate concern. An academic's career is built as a peer in a community of idea-traders; to fail to respect the culture of attribution is to be cast out of that community. Corporate espionage and patent violation aside, there isn't really a culture in the corporate world of citing the source of one's ideas; nobody cares if Lego's inventor was inspired by God or by ten years of cobbling in their workshop as long as the pieces click together. So it is, perhaps, a fair criticism that the creators and consumers don't particularly care about.
And, to come back to the example of building the search index for a Google, it's not an issue for that use case because creators feel sufficiently compensated when they get linked to (even though, technically, we have no idea the pedigree of the search engine's choice to synthesize a link-out... It may have directly crawled the site, it may have decided the site was worthwhile from in-links without crawling it, it may be a manual override or a link harvested from a trusted authority on "cool links." Again, nobody complains about "pedigree" or "plagarism" here unless the system is perceived to not be working to their best interests... Notably, other indexes do complain that Google harvests their link-aggregations without consent!).
Add comment