The AI Threat to Open Societies

> 2. free access to articles → high quality journalism goes out of business as they have to compete against zero-cost alternatives → misinformed (or even deliberately disinformed) society → people less likely to fight for openness.

This can be right in the capitalistic framework many have to live in and deal with. On the other hand this encourages people to write what is sold best / for the highest price.

You can see how damaged the scientific publication system is as organizations and individual scientists are rated by how many articles (quantity not quality) they publish and where they are published (artificial reputation vs. real value).

I am convinced that everything must be open to really pave the way towards an open society.

The problem with this ATM is just the capitalistic framework. Peoples sole purpose in life is earning enough money to not perish before kind of a natural death occurs [which is also connected to amount of money you have].

The need to enslave yourself without purpose just to exist and the reality that success can be easily achieved just by making money somehow (quality doesn't matter and in most cases is punished because you don't earn enough margin) does not foster quality in any way.

Same goes for writing in my opinion. Better to read something from someone who is really dedicated and interested in the topic than some mercenary writer.

The moment complexity on the web goes up, your need for governance structures goes up and the web becomes a force for centralized control of entities on the web: Human or otherwise.

The higher the level of complexity, the better the tools required to manage and analyze data. Therefore the better the tools available to analyze and manage humans.

I guess, if you project this, then the highest levels of the web probably are fully centralized and firewalled networks,.

At some point you have to deal with attackers, inimical and hostile networks and attacks to take over your "stack" of complexity/society. With the internet and high enough complexity, You can finally attack some subset of human behavior, privacy, brains, or information with automation.

The internet is probably the equivalent of flight based damage to fortresses.

And this is before people discuss something like singularity AI.

The acid test for self driving cars is whether they can drive down a road in India or any other country where people don't have western stands of driving discipline. The worse the better.

HOWEVER, And this is the idea I would pursue and sell to authoritarian governments everywhere you don't need Level 5 autonomous cars to get autonomous cars.

It may be abhorrent to the societal values of most of the people reading this comment, but its great to MANY people in China.

But even that, is a red herring.

Strongly controlled and moderated communication networks are currently less brittle than open societal systems.

Even on a micro scale, on forums, you can see the ideas the Chinese apply being put to use out of sheer necessity.

When you Add in state level actors, with the ability to crunch the numbers, then its entirely possible to create the impression of a consensus in human minds with the use of sock puppets/pseudo-human accounts.

The holy grail of such research is discovering intent of speakers or groups of speakers. This will be used to stop hate speech as much as it will to kill dissidents.

This is a right mess and any tool created to clean it up (other than human effort), is liable to create more problems.