"and in case you're like, yeah, but my real life would just be so much better if my top five challenges disappear. would it? i'd like to point out that if super mario brothers was just jumping over one pit, and then you rescued princess peach, no one would play it." -mark rober
what happens to interest rates if ai has runaway capability growth?
define the interest rate as the natural rate of trading things from the future to the present. so when realizing that the return on investments in ai is enormous, you would think it rises a lot. i want to borrow to build my data center and you want to borrow to build yours. because it's going to be so profitable, that's going to jack up interest rates.
but when ai also creates capital and savings way faster than the economy can absorb them, returns on new capital get competed down. abundance should make money cheap, and rates should fall.
its very interesting how different things get automated at different times, based on what the bottlenecks are. the things that get automated are very deflationary, and the bottlenecks are very inflationary, for a short period of time until the bottleneck itself gets automated. then it itself goes really deflationary.
software goes (went?) first. then (now) compute is skyrocketing, and you have people saying yes, i want to take out tons and tons of debt to buy as much compute as possible. but at some point the entire compute supply chain gets automated, especially with the level of innovation that the new models allow for. that goes deflationary, and you no longer need debt to buy compute. but maybe now i take on debt to buy robot factories, if those aren't automated yet, and so on.
and the whole time, as everyone borrows to invest at once, all that money chases the same scarce physical inputs. none of that supply can scale as fast as the demand hitting it, so the prices spike and we get these weird inflationary effects. usually you raise rates to combat inflation, so with high growth, great investment opportunities, and a lot of contention on resources, what does the fed even do?
anyways, went down a bit of a rabbit hole and thought i'd share.
a company tied to a national priority has a demand floor most consumer startups can never touch. what is the best way to learn how government contracts work?
what is the appropriate discount rate for civilization itself? should we value the welfare of someone born fifty years from now at a fraction of our own welfare simply because they don't yet exist? most ethical frameworks suggest not, that future people's interests deserve substantial moral weight, even if they cannot advocate for themselves. yet it seems that our institutions structurally undervalue their interests.
