Technological unemployment, where unemployment arises from (with some oversimplification) workers being replaced by machines, has so far been temporary as new jobs were created that replaced the old jobs. Structural unemployment has been predicted many times in the past and the present, but has not (yet) occurred.
Historical increases in productivity have led to an increase in overall production. Physicist Timothy Garrett models the economy as a heat engine, with more details in papers linked from Garretts personal page:
In each of the past 40 years for which records are available, a continuous 7.1 Watts has been required to maintain every one thousand inflation-adjusted 2005 dollars of historically accumulated economic wealth (not yearly economic output or GDP). As of 2010, civilization was powered by about 17 trillion Watts of power which supported about 2352 trillion dollars of collective global wealth. In 1970, both quantities were less than half this. In the interim, energy consumption and wealth grew equally rapidly in the interim at an average rate of 1.9% per year.
Dr. Garrett is not an economist by training, and I don't know how his models are received by mainstream economists.
How do contemporary forecasts on technological unemployment take into account an economy that is increasingly limited not by the availability of (sufficiently skilled) labour, but rather by the availability of resources?
To illustrate it further: some speculate that in the next years and decades, self-driving cars will displace professional drivers (taxi drivers, lorry drivers, etc.). This will initially increase unemployment, but it would also lower the labour component of the price of many goods. Historically, civilization has responded to such events more with an increase in production than with an overall decrease in hours worked. However, production cannot be increased arbitrarily; resources become increasingly scarce. Therefore, those prices will go up. That means that perhaps this time around, the feedback preventing structural, technological unemployment does not apply... hence my question.
I welcome answers that do and answers that do not take into account externalized costs such as damage to natural ecosystems, pollution, and anthropogenic climate change.