4 ms·

I think it boils down to collection and processing of information. Add to that the fact that different actors in the system follow different kinds of goals. The

I think it boils down to collection and processing of information. Add to that the fact that different actors in the system follow different kinds of goals. The real world is very complex and in order to transport information we need to simplify and quantify according to some model we made up. Further up instances work on the model instead of the real world. Those models become more and more abstract the further up in the chain you go. The reason for this is the limited information processing ability. One human brain can only handle so much. Talking or writing can only transport so much. An organization can only handle so much. We can have worse or better abstractions and less or more efficient organization structures and communication methods. But if your basic unit of information processing are humans with their limited brains it will always be difficult to do any central planning. Especially if the information gets distorted on the way and rules are not followed as they are intended. Then you need to add control systems that take further resources and control systems for the control systems and so on. I assume this is why central planning has failed in so many areas. But what if we could create a super brain that can handle all the detail information collected at the lowest level and integrate it into one huge model of the state of the world? There would be no misaligned incentives inside of this brain, no communication bottleneck, no abstraction loss. I'm sure that we won't "simply create" such a central instance and make it work in a single try. But I'm sure that it could work in some areas and produce very good results. Then it's up to our governments and society on how to use this power.

sharpener 4y ago
Efficient proprioception is useless if key behavioural incentives are misaligned with key aspects of the environment. Humans are quite good at creating collective organisations that are delusional, even though some of those organisations have excellent information about the state of their environment, because the invented incentive structures for action favour simplistic short term "maximisation" of fetishised behaviours instead of optimisation for (even) the medium term healthy future. Resource allocation is still looking frighteningly suboptimal right now, and simply empowering that with a "smart" central AI bot following the same patterns probably won't improve much. I'm skeptical about the human race getting past this. Humans want a lot of stuff, and are prepared to form huge organisations to get that stuff, without really thinking through whether they really need it, what the other impacts will be, or why they are even doing what they do. What is the human race trying to be? Why? How does the pattern of incentives get aligned? It seems like we only have the weakest responses to these questions right now.
- threatripper 4y ago
  Those are important questions we seem to desperately avoid as a society. Things are happening despite not having those answers. My guess is that humans will be sidelined as AI is developing exponentially. There probably won't be a big war human vs machine instead we will probably see humans being turned into some kind of zoo animals as we already are if you look from far away. Then the AI will do its thing in space while we sit here happily staring into the stars.

sharpener 4y ago

Efficient proprioception is useless if key behavioural incentives are misaligned with key aspects of the environment. Humans are quite good at creating collective organisations that are delusional, even though some of those organisations have excellent information about the state of their environment, because the invented incentive structures for action favour simplistic short term "maximisation" of fetishised behaviours instead of optimisation for (even) the medium term healthy future. Resource allocation is still looking frighteningly suboptimal right now, and simply empowering that with a "smart" central AI bot following the same patterns probably won't improve much. I'm skeptical about the human race getting past this. Humans want a lot of stuff, and are prepared to form huge organisations to get that stuff, without really thinking through whether they really need it, what the other impacts will be, or why they are even doing what they do. What is the human race trying to be? Why? How does the pattern of incentives get aligned? It seems like we only have the weakest responses to these questions right now.

threatripper 4y ago
Those are important questions we seem to desperately avoid as a society. Things are happening despite not having those answers. My guess is that humans will be sidelined as AI is developing exponentially. There probably won't be a big war human vs machine instead we will probably see humans being turned into some kind of zoo animals as we already are if you look from far away. Then the AI will do its thing in space while we sit here happily staring into the stars.