Quote (El1te @ Nov 13 2024 12:19pm)
I agree on most of this. I am overall making a distinction between automation and AI, which are related but not the same.
Automation certainly invalidates many jobs - however, it doesn't invalidate human capital - the nature of the jobs changes, people need to find new jobs & learn new skills, etc. For example, many people who learned to code for the world of the 2010's were invalidated by further advances in computational technology, thus thousands upon thousands of them got laid off. Those people will now have to learn a new skill to made ends meet, be that learning to mine, learning to plumb, learning to pipefit etc.
To your line 2), I disagree that the advances are exponential - we have already met the physical limit on transistor size without encountering non-negligible quantum tunneling issues, so computational power is now relatively static. It's not getting more powerful. We can build bigger and bigger computers, but the transistor boom is over. Quantum computers are still largely a fictional imagination, much like nuclear fusion energy - I wouldn't count on these being commercially available in our lifetime.
I also believe it isn't wrong to compare our present situation to the past. The textile mill, steam engine, computer, all technological advancements led to many many people's jobs being supplanted/invalidated. But that didn't mean jobs went away altogether - they simply changed, with perhaps more service jobs becoming available. The idea that people can automate everything & not work is largely a utopian fiction - people 200 years ago thought that and they were wrong, same with people today. The path though as you mention might be bumpy, it might be mildly bumpy or very bumpy, it's hard to predict.
Note that my perspectives are from the perspective of a physical scientist, not a mathematician or computer programmer. I don't code. My brother however is a master coder & mathematician who works with the world's leaders in the field of mathematical programming, optimization & machine learning.
im gonna focus in on the bolded because it's important to understand. the past almost NEVER had full automation. even when machines were created to seam baseballs or weave cloth, human hands were required in many places along the process.
this is not going to be manufacturing in the future, and transportation is heading the same direction. if we lose those industries and humans do zero or almost zero in them, especially with an increasing population, there simply aren't enough jobs left even if emerging industires grow.
a lot of times people rest on the assumption that "well, we can't have 50%+ unemployment, supply and demand. if there's that much supply of workers we MUST find something for them". this is a BRAVE assumption, and not bore out in reality imo.
in the past workers displaced in large areas could move a block over or even a city over and almost instantly find new work. what happens in large swaths of the country where the gap between unskilled and skilled labor is so great that once the plant closes you're just sunk, forever. no options, learn to code or die on welfare.
my main point in all of this isn't that it 100% will come soon, or be as bad exactly as i say, but rather that the chance is so great we need to act preemptively to create systems to help smooth out the bumps. i still think 100 years from now we in america on average will live in 1 bedroom govt apartments and take a stipend which goes up if you volunteer to get yourself spayed or neutered. im taking steps with my family to guarantee that wont be my kids or their kids, we as a country should do the same. instead we sell of most of the land to vanguard and blackrock.