> Michael Schumacher’s family are planning legal action against a German weekly magazine over an “interview” with the seven times Formula One champion that was generated by artificial intelligence.
It does bring up an interesting question about privacy: can stuff made up by a piece of software violate somebody's privacy?
My hot-take of an answer is: yes, probably, depending on the context it's used in. For example, in this context, and even though no specific private information was shared, it really feels like a massive invasion of privacy.
And it's literally taking away agency from Schumacher and his family. Normally you can turn down an interview request. Not here.
@rysiek i know “who thought this was a good idea?” was a rhetorical question but i have to note that this dumb fantasy of digital immortality via a deus-ex-machina AI scanning the history of your fucking social feeds and interpolating and simulating your true essential personality in a Dark Mirror digital heaven has a long history in the dumb AI woo-woo community— eg it’s the basis for Roko’s Basilisk and other longtermist billionaire afterlife thought experiments https://www.vice.com/en/article/evkgvz/what-is-rokos-basilisk-elon-musk-grimes
Somewhat nitpicky answer: yes, if it has negative social consequences for the target (imagine very sarcastic praise).
I'm curious about your opinion on whether slander can be considered a privacy issue, because I think that the source of harm here is more similar to source of harm in slander than in privacy violations (not just typical privacy violations, but the way I'd delineate privacy violations).
@rysiek Yup. I think that trying to put it into those categories harms the future ability to speak precisely, which I value perhaps more highly than people do on median.
@rysiek I also wonder how much the existence of harm depends here on the way society is structured.
For example, I can't imagine a society where gaslighting is not harmful to humans (and maybe even to ~any entity that's curious). Any specific kind of slander is very society-specific (it needs to actually harm one's reputation to be slander). Various ostracism-adjacent actions are harmful in some societies, but nearly neutral in others.
Curiously, I find it easier to find examples of things that are not harmful in the society I live in but are harmful in some society I just read about than v.v.
@trebach@jens sure, and there is a respectful, reasonable way to do such things.
Having a chatbot hallucinate an interview with someone without their permission and without the permission of their family is neither respectful nor reasonable.
@jens@trebach I think the Monkhouse example at least gets pretty close here: it's done with the explicit consent of the family and in order to raise awareness about an issue that can be reasonably assumed to have been important to him.
Add comment