"What Silicon Valley calls “artificial intelligence” at the moment is something more like the computer aboard the USS Enterprise, which can respond to voice prompts and answer a wide range of queries instantaneously. Where is the nearest solar system? How many Klingon warships are pursuing us? How many life-forms are on the planet below? These answers can come much more quickly than any human could give, but they also do not reflect creativity or imagination in any meaningful sense...
Randomized and uncontrolled generation renders the result ineligible for copyright, the office said. While it noted that Kashtanova said she went to great lengths to enter prompts specific enough to produce her desired result, that input was not the same as actual creative work. The office compared her to a patron who commissioned a piece of artwork from a client based on specific instructions. If she had given the same instructions to a human artist that she gave to Midjourney, the office noted, Kashtanova herself would not be able to claim the copyright of the piece that the artist ultimately produced.
That explanation is a laudably insightful understanding of labor and creation by the Copyright Office. It may be even more apt than the government realized. A “generative A.I.,” after all, is only as good as the material it is “trained” on—the corpus of raw text or images that the algorithm then uses to produce simulacra of new text and images."
No comments:
Post a Comment