>That's not at all true, or at the very least, it requires a raft of other assumptions which may not hold, to all stay true, in order for the consequence to sta

>That's not at all true, or at the very least, it requires a raft of other assumptions which may not hold, to all stay true, in order for the consequence to stay true. Like for example the cost of rare earth metals, good trading relationships between nations, and so on. Sure, those lines aren't monotonically decreasing/increasing but the trend isn't likely to reverse direction any time soon. >I'd agree, but only for a very narrow set of "solutions". Scaling something like twitter is just in no way whatsoever equivalent to scaling something like self-driving cars. There'll definitely be a costly hardware buy-in for a lot cases, but some applications of AI might only need a camera - say, a replacement for a cashier at a grocery store. >There is a risk, but there are risks of many things happening. My guess is that we're not going to handle this. Just as in software development we have the YAGNI principle, I don't think it's a good idea to make policy based on future predictions that are very cloudy at best. Sure, there's a risk it will happen, but there are actual problems here today demanding resources and time... I think it's worth discussing because waiting for the bad outcome to happen before trying to handle it could be too little too late. Even if the percentage risk of an economic shock is low, the magnitude of the potential damage is high. The average pre-tax income of the bottom 50% has stagnated since 1980 at about $16,000 per adult [1] These people are more likely to live paycheck-to-paycheck and more likely to work jobs which might be targeted by automation. I think it's a recipe for disaster. [1] http://eml.berkeley.edu/~saez/PSZ2016.pdf http://eml.berkeley.edu/~saez/PSZ2016.pdf