> Robotics production, tooling, software engineering, and the jobs associated with whole new industries that are enabled by robotic manufacturing.
All of which employ far fewer people (And collectively pay them far less in wages) then the assembly lines did. If they didn't, the robots would be more expensive then assembly line workers.
> Also, the jobs associated with more cars being on the road and being more reliable.
More reliable cars arguably produce fewer jobs. More importantly, per-capita car ownership hasn't changed in the US in the past 35 years.
> More broadly, economists overwhelmingly think that increases in productivity lead to improvements in quality of life of people over time. Either by improving their labor situation (better, safer jobs) or their consumer situation (cheaper goods, more choice etc).
Have these economists ever considered that automation scales non-linearly? In the industrial revolution, if you wanted to double the amount of widgets you made, you needed to double your workforce. If you want to double the number of widgets you make today, you buy more robots - this does not double your costs.
If we take automation to its logical conclusion (Wherein robots can build other robots), then its clear that that claim is nonesense. Yes, the labor situation is improved (All jobs are safe, because no jobs are left), and yes, the consumer situation is improved (All goods are now made by robots, all robots are now made by robots). Yet, it's a complete disaster - as with no employees, nobody has the money to buy all these cheap goods.
For a capitalist, the best state of affairs is when he doesn't have to employ anyone to produce products. If automation is adopted, it's because it gets a factory closer to that point.
(And let's not forget that exponential growth in the production of goods will bring about an unmitigated ecological disaster... We need to produce fewer things - not more.)