David - nice post. It's fun to think about what changes will come and how much we've changed. I think AI will replace a huge amount of jobs over the next 5-10 years (replace or make people so much more efficient, less workers will be needed): driving (and other transportation), medicine (make Drs really efficient), financial planning, lawyers, manufacturing, architecture, engineering, ... - it's hard to think of an industry that would be untouched - it may require a universal minimum salary - it could really disrupt our society - because of the cost, maybe a country like Mexico would lag in it's implementation - we'll see
If AI becomes so transformative, I imagine the U.S. government would appropriate it, or regulate it heavily as a matter of national security. I assume adoption is likely to be slow—technology might advance quickly, but people and institutions don’t. In my neighborhood, some folks still use donkeys to haul firewood up the hill, despite countless better fuel options from the last few centuries.
But longer term, I wonder if Zvi Mowshowitz is right (reacting to the recent Citigroup report on AGI in 2029):
“All most people can do is:
1. Save as much money as possible before then (ideally invested into things that do well during the transition)
2. Become a citizen in a country that will have AI wealth and do redistribution”
There are a lot of pessimistic takes about the future but I share your optimism. I hope they invent cold fusion and it puts me out of a job in solar. We’re making good progress against climate change. The world is a better place than it was 25 years ago and I hope, and expect, it will be a better place 25 years hence.
Yes I’ve got very mixed feelings. I’ve read that author’s article in asterisk magazine. I’ve actually run some of his numbers through some of my financial models and I’m not sure that I agree with a number of his assumptions but what he’s proposing is likely *directionally* possible.
David - nice post. It's fun to think about what changes will come and how much we've changed. I think AI will replace a huge amount of jobs over the next 5-10 years (replace or make people so much more efficient, less workers will be needed): driving (and other transportation), medicine (make Drs really efficient), financial planning, lawyers, manufacturing, architecture, engineering, ... - it's hard to think of an industry that would be untouched - it may require a universal minimum salary - it could really disrupt our society - because of the cost, maybe a country like Mexico would lag in it's implementation - we'll see
If AI becomes so transformative, I imagine the U.S. government would appropriate it, or regulate it heavily as a matter of national security. I assume adoption is likely to be slow—technology might advance quickly, but people and institutions don’t. In my neighborhood, some folks still use donkeys to haul firewood up the hill, despite countless better fuel options from the last few centuries.
But longer term, I wonder if Zvi Mowshowitz is right (reacting to the recent Citigroup report on AGI in 2029):
“All most people can do is:
1. Save as much money as possible before then (ideally invested into things that do well during the transition)
2. Become a citizen in a country that will have AI wealth and do redistribution”
There are a lot of pessimistic takes about the future but I share your optimism. I hope they invent cold fusion and it puts me out of a job in solar. We’re making good progress against climate change. The world is a better place than it was 25 years ago and I hope, and expect, it will be a better place 25 years hence.
One of my favorite Reddit/Quora/Perplexity questions is “how would humanity change with limitless energy.” Terraforming the West intrigues me, though I imagine you have mixed feelings with your newfound love of the desolate desert https://caseyhandmer.wordpress.com/2024/10/26/we-can-terraform-the-american-west/
Yes I’ve got very mixed feelings. I’ve read that author’s article in asterisk magazine. I’ve actually run some of his numbers through some of my financial models and I’m not sure that I agree with a number of his assumptions but what he’s proposing is likely *directionally* possible.