which could get especially weird with AI-powered election interference
Something to clarify here: AI interferes with people in general, not just elections. There is a ludicrously wide variety of ways that AI-powered interference can cause haywire, notably including but not limited to targeting people who don’t use social media. I’m particularly interested in systems that target specific elites and compromises networks, not just berlin-wall-fall-style erosions of regime legitimacy in public opinion that pulls out the rug from under them.
My primary threat model here is of the form “Trump (or a Trump successor) attempts another bureaucratic coup, but this time succeeds.” I am particularly worried about this because I think it would very substantially undercut the possibility of international cooperation on AI governance, as I mentioned here.
My model of that was that if a Trump faction took control (assuming that such a faction existed and was capable of doing that, as opposed to the default explanation of it being not even the top 10 political PR stunts in history), history would have been written by the victors and constitutional rule probably would have continued by Trump and the democrats accusing eachother of stealing the election. It wouldn’t have been great for AI governance, but it probably wouldn’t be nearly as impactful as transformative paradigm shifts involving the security services, such as AI-powered IoT surrounding every government building and penetrating most homes, corporate/executive offices, and conversation zones like restuarants.
Worrying about AI governance is of course the right approach here, because the enemy’s gate is down and AGI is the finish line for most things. I’m just disagreeing on the details of whether political organizations or security organizations are more relevant in this post’s particular context.
Something to clarify here: AI interferes with people in general, not just elections. There is a ludicrously wide variety of ways that AI-powered interference can cause haywire, notably including but not limited to targeting people who don’t use social media. I’m particularly interested in systems that target specific elites and compromises networks, not just berlin-wall-fall-style erosions of regime legitimacy in public opinion that pulls out the rug from under them.
My model of that was that if a Trump faction took control (assuming that such a faction existed and was capable of doing that, as opposed to the default explanation of it being not even the top 10 political PR stunts in history), history would have been written by the victors and constitutional rule probably would have continued by Trump and the democrats accusing eachother of stealing the election. It wouldn’t have been great for AI governance, but it probably wouldn’t be nearly as impactful as transformative paradigm shifts involving the security services, such as AI-powered IoT surrounding every government building and penetrating most homes, corporate/executive offices, and conversation zones like restuarants.
Worrying about AI governance is of course the right approach here, because the enemy’s gate is down and AGI is the finish line for most things. I’m just disagreeing on the details of whether political organizations or security organizations are more relevant in this post’s particular context.