- The ability of Robots to do things that humans previously did has been slowly drifting up the IQ scale for some time now. Robots, for instance, build microchips, with a couple human operators doing the work that used to be done by lots of human assemblers. While it's true that we need SOME tech support for that, it's also true that ROBOTS + robot operators + tech support means that it requires 6 people to do a job that used to take 100.
- Not all of it is low-IQ work. Some of it is just routine work (Tax preparation is now frequently done by Quicken, rather than a tax preparer).
- The complexity of the tasks that robots/software can do has been increasing for a long time now. Stuff robots used to do poorly is now stuff that robots/software can do well. For instance, there has been measureable increase in quality in automated telephone systems over the last 5 years.
- AI has been around the corner for 20 years for at least 40. I've got friends deep in the field as well. So one should expect AI to remain a forever-unsolved problem? NO. Moore's law has been stable for >100 years, and every 5 years, someone announces a new reason to believe that Moore's law will fail. No luck on either changing yet.
- IF Moore's law continues unabated, we have a question of whether AI is a hard problem, an easy problem, or a medium difficulty problem. FWIW: I define a medium difficulty problem as being: we will get Human-equivalent AI when we have Human-brain equivalent processing to throw at the problem. Standard desktops will, under Moore's law, hit this target near 2040. I count easy problems as one that takes 1-3 orders of magnitude less processing...and hard problems as ones that take 1-3 orders of magnitude more processing. Under Moore's law, we have a range for human-equivalent AIs of 2030 to 2050.
- Even under crazy Cartesian dualistic positions on the impossibility of AI, we'll still get Artificial Stupids as per Michael Flynn, which can do HUGE amounts of work for us.
- Basically, that means that the cost of labor for IQ-100 intellectual work (answering phones) should drop to zero sometime in that range. And when stuff drops to zero cost...you get more of it. Similarly for low-IQ occupations in the physical world (fruit picking, janitorial, dishwashing).
- What we will have then is a labor force that offers (at near-zero cost) ALL of the physical/intellectual jobs that a person with a 100 can do. Farm work and industrial work will be 100% (+/- 3, as per all Aretae claims) automated, and much Service work will as well.
- What happens when the price of labor equivalent to 50% of the labor market is zero (+/- 3%)?
- Material scarcity stops being an issue.
- The idea that work will define how much of your material needs/wants get met also fails.
- Socialism is the only word we've currently got for a system where your NEEDS are taken care of.
The virtue of excellence
Saturday, May 21, 2011
Singularity approaches, part II
My prior post on post-material scarcity generated a bunch of critique. Let me then lay out the position more clearly: