From a cynical point of view, the point of democracy is to get the people to blame each other - for voting for the wrong party, for not voting at all, for voting for one of the two big parties or for voting for the spoiler. And thus prevent revolution or murder of politicians.
Inferring who you would vote for doesn't accomplish this as well.
What if the AI looked at all the data and voted for the representatives who'd be best to better their "master"'s life?
Although then, Zuck's, Bezos' and Musk's AI would still send out lobbyists to manipulate politicians to fuck us plebs over (Hah, the overpaid codemonkey thinks he's a pleb...).
It'll be a future when an AI orders escorts be sent to Supreme Court justices' rooms in the golf vacations paid for by the billionaire. "Data says, Justice [____] likes long haired blondes with nice buttocks, send Tiffany. She should bring beer, he likes beer.".
How does this account for new information and changing environments? In a real-world setting, respondents are constantly updating themselves, whether consciously or not, by interacting with people, their communities, or their representatives.
Unless models are updated by attending town halls, interacting with communities, gaining some kind of lived experience to shape a view point, or something else I haven’t thought of, then I am unsure how robust this approach may be to a changing landscape.
The right state of mind isn't, "what about all these reasons this stupid AI approach is flawed," it's "what will society be like when this stupid AI approach inevitably happens because party A profits from selling it while party B saves money from buying it?"
One step closer to Managed Democracy.
It's an idea from a video game where AI is spying on your life 24/7 and infers who you would vote for so you never actually need to (or can) vote.
From a cynical point of view, the point of democracy is to get the people to blame each other - for voting for the wrong party, for not voting at all, for voting for one of the two big parties or for voting for the spoiler. And thus prevent revolution or murder of politicians.
Inferring who you would vote for doesn't accomplish this as well.
Reminds me of this short story by Asimov with nobody votes anymore just computer predicts the outcome. https://en.m.wikipedia.org/wiki/Franchise_(short_story)
What if the AI looked at all the data and voted for the representatives who'd be best to better their "master"'s life?
Although then, Zuck's, Bezos' and Musk's AI would still send out lobbyists to manipulate politicians to fuck us plebs over (Hah, the overpaid codemonkey thinks he's a pleb...).
It'll be a future when an AI orders escorts be sent to Supreme Court justices' rooms in the golf vacations paid for by the billionaire. "Data says, Justice [____] likes long haired blondes with nice buttocks, send Tiffany. She should bring beer, he likes beer.".
How does this account for new information and changing environments? In a real-world setting, respondents are constantly updating themselves, whether consciously or not, by interacting with people, their communities, or their representatives.
Unless models are updated by attending town halls, interacting with communities, gaining some kind of lived experience to shape a view point, or something else I haven’t thought of, then I am unsure how robust this approach may be to a changing landscape.
The right state of mind isn't, "what about all these reasons this stupid AI approach is flawed," it's "what will society be like when this stupid AI approach inevitably happens because party A profits from selling it while party B saves money from buying it?"
Polling measures where things are. AI approximates where things were.
Somewhat related, the 1947 movie Magic Town.