#AI #GenerativeAI #AIBubble #AIBacklash: "As AI tools become more capable, and get used against more and more people, this distinction could disappear, and there are plenty of contexts in which deepfakes and voice cloning are already used for straightforward deception. For now, though, when it comes to bullshit generated by and about public figures, AI is less effective for tricking people than it is for humiliation, abuse, mockery, and fantastical wish fulfillment.
It’s good for telling the sorts of lies, now in the form of images and videos, that are stubbornly immune to correction because nobody thought they were strictly true in the first place. Swift’s post is typical of the popular AI backlash in that it’s both anticipatory — this stuff sure seems like it could go catastrophically wrong someday! — and reacting to something less severe that’s already happening — hey, this shit sucks, I wish it would stop! It’s a reasonable response to a half-baked technology that isn’t yet deceiving (or otherwise reshaping) the world at scale but in the meantime seems awfully well suited to use by liars and frauds. Her experience is aligned with that of the general public, for whom theories of mass automation by AI are still mostly abstract. For now, they just see the internet filling to the brim with AI slop."
https://nymag.com/intelligencer/article/taylor-swifts-endorsement-is-part-of-the-ai-backlash.html
