How do you stop deepfakes from taking over social networks? OpenAI has some ideas. I spoke with some members of its policy team about how it's thinking about the problem ahead of the election https://www.platformer.news/open-ai-deepfakes-election-c2pa/
@caseynewton the social networks aren't going to do anything about it. They'll pretend to care about it as cover for collecting and selling more data about their users to government and surveillance companies, maybe suggest some legislation to their pet politicians to that effect with a name like "Protecting Our Children In the Digital Commons", pat themselves on the back and get back to the business of making money. And, OpenAI will be both a buyer and seller of that data.
We do not need central control to cope with a problem caused by centralization. We do not need ranked feeds at all. We do not need virality.
PS: OpenAI et al helping to create deep fakes and then selling remedy reminds me of creating virusses and then selling antivirus programs or even ransomware.
@caseynewton That would be a really depressing way to live even if it worked. A pervasive system of digital surveillance, a DRM-locked cryptographic lifecycle from cameras to MS Paint to Facebook posts, a cat-and-mouse race of expensive algorithms all based on statistical plausibility, and downranking people who take a screenshot anyway -- all to try to stop a few people who can't stop using AI for deception and amusement?
It might be cheaper to hire someone to stand over each of us and whack us with a magazine when we reshare AI anyway. The only people these plans would benefit are the companies proposing them, and I think it's a fig leaf over both unsolvable problems with the tech, and companies' unwillingness to slow down and cultivate norms around it.
Add comment