4 ms·

Yep, and China's authoritarian outlook and lack of privacy concerns might allow them to win a future AI race, because there's no western concept of rights. Swee

Yep, and China's authoritarian outlook and lack of privacy concerns might allow them to win a future AI race, because there's no western concept of rights. Sweep up as much data as you want and throw it into models. Track every citizen, every click, every purchase. The US may fall behind simply because we believe in ideas like privacy. Kai-Fu Lee's book goes into this in detail. Although if we're being realistic, less scrupulous companies such as Amazon will step in and take the contracts.

mindslight 8y ago
Or they may fall on their face, like every other attempt at central planning has. The economy is basically a distributed optimization heuristic. If you're talking about "falling behind" in the global sense (using "AI" to centrally plan), then you're hypothesizing that we can essentially port the distributed algorithm into the silicon realm and further optimize it (eg prune destructive adversarial behavior). We certainly have more computing power now than say that past attempt based on fluid flows, but that's still a huge if. If you're talking in the local sense (that "AI" will help companies themselves optimize better), than whatever extra percentage gained from that will not be enough to outrun the State deploying the same technology (as in the above goal), and destroying the individual businesses. One thing I do know is that even if this turns out to be the winning strategy, sacrificing our values in an attempt to compete at it will still only net us second place! -- Having written that, I think this implicit assumption that it will be economically beneficial for AI to be directed at individuals and individual behavior needs to be directly examined. How much individual productivity is actually being left on the table? The US system is quite good at motivating people, so much so that it's common for lower classes to see themselves as "temporarily embarrassed millionaires". There are a few deviants, but for the most part lack of individual productivity is a failing of the larger structure. Heck, most of the US's criminals are in jail for engaging in economic activity! Rather, the narrative seems like the age-old keep-the-plebs-in-line play, not a legitimate approach to economic success. If AI has something to offer for planning the economy, then wouldn't that be properly applied to the government-company interface? That's where the structural gains are actually available to be had!
- ardy42 8y ago
  > Or they may fall on their face, like every other attempt at central planning has. China isn't centrally planned anymore, and hasn't been for a long time.
  - mindslight 8y ago
    Yes, and perhaps they'll even continue on that path if it is really an attractor. But there are two motte-and-bailey terms in the above comment - "ahead" and "AI". If we're only talking about the narrow scope of China being ahead in AI, with "AI" meaning predicting outliers from large populations, then that is not so worrisome! Google (et al) can simply buy the better population surveillance technology from China and apply it to their own business, as the US will still be generally ahead. But rather such talk seems to be aimed at invoking a worry that a lack of individual rights means China will succeed at a more generally useful "AI" (whatever that means), and that it will then be applied to their economy to get them "ahead" in general. This is the underlying idea I am addressing.

mindslight 8y ago

Or they may fall on their face, like every other attempt at central planning has. The economy is basically a distributed optimization heuristic. If you're talking about "falling behind" in the global sense (using "AI" to centrally plan), then you're hypothesizing that we can essentially port the distributed algorithm into the silicon realm and further optimize it (eg prune destructive adversarial behavior). We certainly have more computing power now than say that past attempt based on fluid flows, but that's still a huge if. If you're talking in the local sense (that "AI" will help companies themselves optimize better), than whatever extra percentage gained from that will not be enough to outrun the State deploying the same technology (as in the above goal), and destroying the individual businesses. One thing I do know is that even if this turns out to be the winning strategy, sacrificing our values in an attempt to compete at it will still only net us second place! -- Having written that, I think this implicit assumption that it will be economically beneficial for AI to be directed at individuals and individual behavior needs to be directly examined. How much individual productivity is actually being left on the table? The US system is quite good at motivating people, so much so that it's common for lower classes to see themselves as "temporarily embarrassed millionaires". There are a few deviants, but for the most part lack of individual productivity is a failing of the larger structure. Heck, most of the US's criminals are in jail for engaging in economic activity! Rather, the narrative seems like the age-old keep-the-plebs-in-line play, not a legitimate approach to economic success. If AI has something to offer for planning the economy, then wouldn't that be properly applied to the government-company interface? That's where the structural gains are actually available to be had!

ardy42 8y ago
> Or they may fall on their face, like every other attempt at central planning has. China isn't centrally planned anymore, and hasn't been for a long time.
- mindslight 8y ago
  Yes, and perhaps they'll even continue on that path if it is really an attractor. But there are two motte-and-bailey terms in the above comment - "ahead" and "AI". If we're only talking about the narrow scope of China being ahead in AI, with "AI" meaning predicting outliers from large populations, then that is not so worrisome! Google (et al) can simply buy the better population surveillance technology from China and apply it to their own business, as the US will still be generally ahead. But rather such talk seems to be aimed at invoking a worry that a lack of individual rights means China will succeed at a more generally useful "AI" (whatever that means), and that it will then be applied to their economy to get them "ahead" in general. This is the underlying idea I am addressing.