Great podcast guys!
A couple of points regarding the idea of automation replacing people and its affects on the middle class:
1) Jarod Lanier expressed quite similar thoughts (albeit a bit longwinded) in his book “Who owns the future?”.
2) A lot of seemly trivial manual labour tasks are very difficult to automate. Dexterous unplanned or semi-planned manipulation of objects is still an active area of academic research and very far from solved. I don’t see a robotics platform that can help with general house cleaning, in a capacity like a maid or hired cleaner, any time soon. Sure, a Roomba can (kind of) clean the floor for you, but not as well as a human and requires humans to restructure environments, set up barriers, ensure IR visibility to a charging station, clear its dust repository etc.
One way to think of this if you want to frame it in your terms is that there are a lot of seemingly mundane daily tasks that require a surprisingly large amount of creativity. Placing dishes into a dish washer is another example. We have the dishwasher appliance, which is a robot as it has sensors and acts based on its senses (water temperature, load, timers etc). But there isn’t a robot that can load the dishwasher for you apart from very trivial demos that are either pre-programmed or require all possible dishes to be modelled and placed in the sink so that they are easily sensed. This kind of demo requires a lot of manual labour to set up 🙂
3) Somewhat related, there is also a lot of R&D in the human-robot interaction (HRI, kind of like HCI but with actuators). There are going to be many tasks that require both humans and robots to collaborate, at least in the next few decades. IMO, this will still reduce the number of human workers and lower the skill requirement of the workers; both of which will drive job numbers (or working time) down.
Full disclosure: Part of my time is spent as an Associate Investigator in a Research Centre in Australia working on this kind of thing http://roboticvision.org/.
4) Healthcare is very hard to automate. I work in a bionic eye (cortical implant) research centre and my limited experience with medical devices suggests that the rate at which medical tech. can be updated is going to be very slow. Much slower that what the tech. industry is used to. The amount of testing, especially for anything that interacts with humans in any physical manner (robots, implants, wearables) is very time consuming and onerous (as one would expect).
This may not be as true in some markets, but a lot of the R&D is funded by and targeted at richer countries.
This again meshes with what you two said about “creative” work. A robot or automated system can help with a lot of healthcare tasks or make them more efficient, but well trained human workers are still needed. This is true both in the “service” side of things (nurses, doctors etc) and the tech. development side of things (engineers, neuroscientists etc).
More full disclosure: My other time is spent at Monash Vision Group, a small R&D centre developing a cortical visual prosthesis http://www.monash.edu.au/bioniceye/resources.html.
Wai Ho