As always, it will be used on the least wealthy and politically connected people first, because they'll be in the worst position to defend themselves.
I think we have to assume that "persuasion-at-scale" is already on the way - I mean we know as a fact that it's happening today via political interests astroturfing social media movements. Soon enough your personal AI assistant will start to subtly manipulate you as well, if this hasn't already begun. This will challenge democracy in an entirely new way; if your technology can reliably alter the political beliefs of say 75% of its users in whatever direction you desire, is voting still a meaningful representation of individual interests?
We must assume that in the next decade or two the concept of free will itself could be challenged by increasingly sophisticated and personalized persuasion technology.
Anyhow, wasn't democracy losing already without llms? As in, voters were already voting against their own interests due to social media and media in control of people with agendas.
Idk, I think bullshitting people works just too damn well anyhow. Doesn't even have to be sophisticated.
Consider someone 50 years ago saying: "Anyhow, aren't people already losing their privacy by sometimes buying things with credit?"
The implied message is that things can't get that much worse, that some practical saturation point is about you be reached. But today our daily movements are being tracked by cell networks and license plate readers, and much of what we type or view online becomes individualized dossiers.
I think the same pattern applies here: If it seems like it can't get (much) worse, then somebody's not imagining hard enough.
If we can use that on politicians (not even manipulation, just decoding), things could get interesting in a hurry...
As always, it will be used on the least wealthy and politically connected people first, because they'll be in the worst position to defend themselves.
I think we have to assume that "persuasion-at-scale" is already on the way - I mean we know as a fact that it's happening today via political interests astroturfing social media movements. Soon enough your personal AI assistant will start to subtly manipulate you as well, if this hasn't already begun. This will challenge democracy in an entirely new way; if your technology can reliably alter the political beliefs of say 75% of its users in whatever direction you desire, is voting still a meaningful representation of individual interests?
We must assume that in the next decade or two the concept of free will itself could be challenged by increasingly sophisticated and personalized persuasion technology.
Anyhow, wasn't democracy losing already without llms? As in, voters were already voting against their own interests due to social media and media in control of people with agendas.
Idk, I think bullshitting people works just too damn well anyhow. Doesn't even have to be sophisticated.
Consider someone 50 years ago saying: "Anyhow, aren't people already losing their privacy by sometimes buying things with credit?"
The implied message is that things can't get that much worse, that some practical saturation point is about you be reached. But today our daily movements are being tracked by cell networks and license plate readers, and much of what we type or view online becomes individualized dossiers.
I think the same pattern applies here: If it seems like it can't get (much) worse, then somebody's not imagining hard enough.