Should machines replace human workers?

Would I like to work for that little? Would I like to compete with robots? Would I like to give up my stuff? You are asking me these questions as if there's a choice. As more people lose jobs, more people will be willing to work for less. The cost of automation is only justified as long as people are a more expensive alternative. If people can undercut automation, they will be hired.

Your second point is completely valid and I agree. I assumed this was topic was operating under the assumption that everything should be automated, including those newly created jobs.

There are a few problems with your thinking.
1) It assumes there is a massive portion of the people willing to work for dirt cheap and be poor. Very poor. I don't think that is the goal of any modern society.
2) It assumes that there is some kind of permanent labor shortage that will only grow. This is most definetly not true. As new technology comes out it creates more jobs then it kills, then in the long run it kill more then it creates, but then more technology comes out that again creates jobs.
3) It would kill GDP. Just look at China were there is very little automation. There GDP per person is around 8400 compared the the US of 48k.
http://en.wikipedia.org/wiki/List_of...PP)_per_capita
Meaning the average US worker produces 8k as much stuff, and should have about 8X as much stuff (varying dependant on distribution of wealth in each country). Are YOU willing to gives up over 80% of your stuff just so there is no automation? I think the answer is a resounding no.

In certain countries it doesn't make sense to go from no automation (low GDP, low wages) to fully automating everything. So I agree with the example of South Africa (11k GDP per person). Yet, as GDP grows, wages tend to grow, and some sort of automation becomes necessary. Take for example the US. Minimum wage here is $7.50. If the lawnmower costs $120 and it takes 1 person 8 hours to mow this field, your talking $60 a day in wages. Three people would cost $180 a day, literally the lawnmower would pay for itself in a single day.

Suppose all jobs are done by robots, or at least enough so that the ratio of jobs:people is tending towards 0.
Does it still make sense to talk about GDP?
Does it even still make sense to talk about any kind of money?
For example, when I buy a loaf of bread, what am I paying for?
I'm paying for 1. people to grow and harvest fields, 2. people to work some kind of mill, 3. people to work some kind of oven 4. people to distribute the bread to a shop near me, 5. people to work in the shop.
If these jobs are all automated by robots what am I paying for? nothing.
That being said, I'm no student of finance or economics so I wouldn't know for sure if you would still put any value in the loaf but I'm pretty sure most people wouldn't pay for something that cost nothing to produce and distribute.

As more people lose jobs.....

...
The fact is that - this "Population growth vs Technology" argument has been disproven by 100 years of evidence. We use more machines today - AND the population is significantly larger than when this argument was first proposed....

I think what this discussion (I hesitate to call it an argument) boils down to is will/when will the Jevons paradox expire? Mechanisation and automation has the effect of boosting productivity per worker but due to increased demand not changed overall employment too much. So agricultural mechanisation may make X% agricultural workers redundant but jobs in the factories to build the tools go up ~X%. Obviously this isn't necessarily good for the individual who looses their job (and possibly their farm and home) because they may find factory work unfulfilling, may not be qualified/may be over qualified or might not be located close to said factory.

But the discussion still occurs because we haven't yet answered the question of what to do if at some point mechanisation/automation starts to overtake demand accross a significant section of the economy (i.e. the labour needed to satisfy the extra demand created by increased productivity and support industry does not meet the labour made redundant). Nor do we really know if that will happen or if it couldn't happen.

IMO this discussion will continue indefinitely unless we do start getting to the point where there simply aren't enough jobs to go around in an economy that can (productivity wise) easily meet the demands of everyone.

The design, testing, production, installation, operation, maintenance, upgrades, retrofits, and replacements of machines all (potentially) require human labor or supervision. It's hard to imagine a scenario where humans are completely replaced.

It doesn't have to be completely replaced, problems arise when the number is reduced to significantly less than that of people able to work. As I said though its a discussion seemingly without end unless it happens.

Certainly, the unmanned drones come to mind.

I heard a comment recently that UAVs are in a similar position to aircraft in the 1920s, shown good promise in combat and now entering the realm of civilian uses (in a myriad of ways). Given the huge scope of potential tasks it will be interesting to see how UAVs will be utilised in the near future.