I'm analyzing what happens to the US economy in the short-term
aftermath of the typical job being replaced by AIs and robots. Will
there be a financial crisis? Short answer: yes.
Let's say 50% of workers lose their jobs at the same time (around
2030), and they're expected to be permanently unemployed. (I know this
isn't fully realistic. I'm starting with simple models and will add
more realism later.)
I'll assume that AI starts making the world more productive around the
same time that this job loss occurs, but that big innovations such as a
cheap cancer cures or the ability to conquer the world are still far
enough in the future that financial markets aren't ready to price them
in.
These assumptions are designed to help me analyze the effects of job
loss with minimal complications from other effects of AI. I'm focused
here on the short-term financial and political consequences of job
losses. There will be some radically different longer-term consequences,
but I'm only analyzing those here to the extent that I expect markets
to reflect them at the time of the job losses.
This post is merely an outline of what a rigorous analysis would look
like. It's good enough for informing my investment strategies, but not
for persuading politicians to adopt better policies.
Note that this will be one of my least readable blog posts. Most of you
should start by reading the conclusion, and only reading the rest if
you're tempted to argue with my conclusions.
If you still think my conclusions are wrong, you can find some more
detailed explanations of my reasoning in this conversation with
Gemini.
Note that I'm targeting this at readers with a significant background
in finance. Please question the details of my analysis, and produce
competing guesses based on answering similar questions.
Conclusions
I expect turmoil similar to that of the pandemic. My median guess is
that it will be somewhat less sudden than the crash of March 2020, and
that markets will mostly recover in one to two years (assuming we have
years before something more dramatic happens).
The financial turmoil is likely to induce political instability. I find
that hard to predict.
The US government will need to be more competent than it was during the
pandemic in order to avoid hyperinflation or defaulting on its debt.
The magnitude of the turmoil will likely be heavily influenced by
hard-to-predict expectations.
Maybe a bright spot is that a financial crash could slow capability
advances at roughly a time of near-maximum risk related to AI alignment.
But that might be offset by politicians being too distracted to do
anything competent about alignment.
I'm surprised at how much my outlook fluctuated while writing this
post, between optimism and despair, before settling on an intermediate
mood.
The process of writing this post convinced me to (slowly) start selling
my remaining (small) positions in bank stocks. I'll be less willing to
sell my stocks in gold mining companies. I'll probably be more willing
to sell some of my other stocks when I've guessed that they've reached
bubble levels, rather than hoping to sell close to the peak.
I'm analyzing what happens to the US economy in the short-term aftermath of the typical job being replaced by AIs and robots. Will there be a financial crisis? Short answer: yes.
This is partly inspired by my dissatisfaction with Tomas Pueyo's analysis in If I Were King, How Would I Prepare for AI?.
Let's say 50% of workers lose their jobs at the same time (around 2030), and they're expected to be permanently unemployed. (I know this isn't fully realistic. I'm starting with simple models and will add more realism later.)
I'll assume that AI starts making the world more productive around the same time that this job loss occurs, but that big innovations such as a cheap cancer cures or the ability to conquer the world are still far enough in the future that financial markets aren't ready to price them in.
These assumptions are designed to help me analyze the effects of job loss with minimal complications from other effects of AI. I'm focused here on the short-term financial and political consequences of job losses. There will be some radically different longer-term consequences, but I'm only analyzing those here to the extent that I expect markets to reflect them at the time of the job losses.
This post is merely an outline of what a rigorous analysis would look like. It's good enough for informing my investment strategies, but not for persuading politicians to adopt better policies.
Note that this will be one of my least readable blog posts. Most of you should start by reading the conclusion, and only reading the rest if you're tempted to argue with my conclusions.
If you still think my conclusions are wrong, you can find some more detailed explanations of my reasoning in this conversation with Gemini.
Note that I'm targeting this at readers with a significant background in finance. Please question the details of my analysis, and produce competing guesses based on answering similar questions.
Conclusions
I expect turmoil similar to that of the pandemic. My median guess is that it will be somewhat less sudden than the crash of March 2020, and that markets will mostly recover in one to two years (assuming we have years before something more dramatic happens).
The financial turmoil is likely to induce political instability. I find that hard to predict.
The US government will need to be more competent than it was during the pandemic in order to avoid hyperinflation or defaulting on its debt.
The magnitude of the turmoil will likely be heavily influenced by hard-to-predict expectations.
Maybe a bright spot is that a financial crash could slow capability advances at roughly a time of near-maximum risk related to AI alignment. But that might be offset by politicians being too distracted to do anything competent about alignment.
I'm surprised at how much my outlook fluctuated while writing this post, between optimism and despair, before settling on an intermediate mood.
The process of writing this post convinced me to (slowly) start selling my remaining (small) positions in bank stocks. I'll be less willing to sell my stocks in gold mining companies. I'll probably be more willing to sell some of my other stocks when I've guessed that they've reached bubble levels, rather than hoping to sell close to the peak.
See my blog for the full post.