By the time AGI is developed, it would be able to improve itself/build improvements, causing the runway effect that leads to superintelligence. The idea of making "personal digital assistants" is humans thinking they can tame a god to be a servant. Taming a superintelligence would be like a dog taming human civilization, and perhaps the gap between man and ASI is larger than the gap between dog and man.
After Brexit, I'm not sure I and many others would be comfortable allowing "greater citizen involvment" such as directly voting on bills or policies. I am politically against such a motion and am a proponent of greater technocracy; letting experts decide what policy is best rather than the masses who can easily be duped by demagogues or other post-truth elements.
This. I'm not sure if the other user means better experience with the services Google et al. offers or if he directly means how Google/Facebook/etc. cater ads/search results/friend suggestions to your web behavior. If it is the latter, then I personally want none of it and I get creeped out in much the same way as meeting a stranger who knows a little too much about me.
There is no such thing as complete rational people, every person goes through bouts of rationality and irrationality, some are just more often rational than others. With that said nobody is immune from irrational emotion sometimes, for example have you been angry from something that wasn't worth it?
The downfall of Afghanistan and Somalia was that the government was too weak. The downfall of Yugoslavia was the death of its benevolent autocrat that held the multi-ethnic society together.
Having an autocrat in and of itself isn't inherently a terrible outcome. Having a malevolent, corrupt, misguided, or sadistic autocrat is what is terrible. A benevolent autocrat on the other hand presents a viable or superior alternative to democracy.