I have sometimes mused about how we have worried about "sky net" since the 90's, but we've already let AI have power to control our lives to some extent.

For instance, there are cases you can point to where some normal person on Twitter with a couple dozen followers posted the wrong thing, it blew up into an international story, and they eventually got fired from their real-world job for it. Who made the decision that this person's views should be widely scrutinized? In a lot of cases it was the algorithm: for whatever reason that tweet drew engagement, so a massive spotlight was put on this mostly-private individual.

In a real way that's an example of an AI system deciding to take away someone's livelihood.

> Who made the decision that this person's views should be widely scrutinized?

This is a serious stretch. The algo didn’t force that person to sign up and post their every ignorant thought on a public forum. People can just put down their phones but they love the attention. Individuals are just as guilty for giving power over themselves to social media.

I'm not "blaming the AI" for this person's behavior. My point is that there was a non-human component in this decision process which led to this person losing their livelihood.

To make an old-world analogue, in 1980 for someone to be held to account for an objectionable view, it probably would have involved some journalist making the decision to write a story about the comments which someone made. That human component in deciding what is newsworthy has been replaced by a machine in some cases.

Another example would be dating apps. Now we have algorithms which make decisions about which parts of the dating pool an individual will have access to. In a real sense, non-human intelligence is playing some role in who gets to pass on their genes to the next generation, and with whom.

I'm not saying that our lives are ruled by machines, but they're not completely not governed by machines either.

It's not the ignorant thoughts which are considered wrongthink nowadays, but reasonable opinions that don't toe the line on controversial topics.

If anything, the next generation of people will learn to avoid online polarized engagement using their own identities. They will create these curated and sterile personas under their real name while using a nom de plume for actual content.

I'm kind of surprised that aliases have not widely taken off already. But I suppose the consequences are not very severe yet.

>I'm kind of surprised that aliases have not widely taken off already.

The funny thing is that these aliases already existed, but we've largely eschewed them for our "real" identities. Prior to probably Facebook I think every interaction I had with other people online was behind a screen name detached from my real world identity.

Aliases used to be the default before the social media revolution, though. If anything the trend has been decidedly in the other direction - using one's real name was seen as irresponsible just 15 years ago, whereas now it seems the average internet user is not even aware pseudonymity is an option.

It reminds me of Spirited Away and how everyone fed No Face for a little gold in return, but in the end they all got eaten.

Wait until people figure out how much power the AIs already have over election outcomes.

What?

Most people in democracies are entrenched into two blocks (even in European multiparty systems), and AI/ML/statistical methods/mathematical optimization allows interested groups to target the people inbetween the blocks. And based on their personality profile you can goad them to one side or the other with targeted advertising.

(I’m not making any claims about where or how or how much these things are happening.)

I'm French and this comment really doesn't reflect our political landscape.

The AI determines which political articles and posts you see. The AI determines if you get reminded to register/vote. At the extremes the AI has a very strong impact on whether you believe Obama is a secret Muslim or Trump is a Russian asset.

Which means people best able to play people to increase the "this post is good" score of the AI will sway the election results. It's not enough to control them, but I'd imagine you can tip a close election by 10% or so.

Dr. Robert Epstein has done some research on this and has testified before Congress about it. Here's one article. https://www.latimes.com/politics/la-na-pol-google-search-bia...

Not reality, but definitely the media we consume. It's no wonder those Search results skew left when most news articles do too.

"Objective absolute indifferent thing, has bias towards patchy intersubjective historically situated thing" sounds a lot less catchy when you unpack it.

It's a joke. Mainstream right-wing socioeconomic theories are falsified by reality more often than mainstream left-wing socioeconomic theories. (Of course, left-wing v.s. right-wing is a bad way to judge a socioeconomic theory; better to check whether it's right.)