On Fri, Dec 7, 2012 at 9:31 PM, Russell Wallace <[email protected]> wrote: > On Sat, Dec 8, 2012 at 2:07 AM, Matt Mahoney <[email protected]> wrote: >> The abstract doesn't mention AI. I estimate the cost of automating the >> global economy, i.e. replacing all human labor. > > Except replacing _all_ human labor is neither possible nor necessary.
That's a good point, because we don't actually replace all human labor. As we replace the simple, repetitive tasks, we create new jobs that are harder to automate. >> I stated the assumption that doing the work of 10^10 people will >> require computing power equivalent to 10^10 human brain sized neural >> networks. > > Sure, unless it doesn't. A lot of human jobs don't actually need full > human intelligence. That's true. A lot of that work has already been automated. But eventually you get to the point where all of the jobs that are left (including the new ones) do require full human intelligence. > Conversely we might end up using more than a human > brain's worth of processing power on a lot of problems, maybe because > we haven't had millions of years to optimize the algorithms, maybe > because we are dealing with exponential search problems that have > unlimited appetite for computing power, maybe because the standard > required to be competitive will keep going up as more computing power > is available. The truth is we don't know how much computing power will > be involved. Computers are already doing a lot of work that humans could not do. >> Nobody has made a molecular scale self replicating robot. Perhaps you >> can explain why such a thing would be impossible. > > Even the hard-core 'nanotechnology will be magic fairy dust' advocates > these days are acknowledging such a thing is unlikely to be developed > because it would cost a fortune while being of little use. I don't think it would be of little use to be able to put each atom exactly where you want. You could make anything at all. By making copies of themselves, it becomes cheap to make even large objects that are precise to the atomic level. One application would be to go inside your body to repair damaged cells to stop aging. Another would be molecular scale computers which are a million times more energy efficient than silicon. Already we can synthesize DNA to create arbitrary proteins that fold into complex shapes of our choosing. Sequencing and synthesis costs are dropping exponentially at a rate similar to Moore's Law. -- Matt Mahoney, [email protected] ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
