Nice post, I agree with the broad point. Thanks for writing!

I think I disagree with the claim [regarding the expert sample of economists] "I think they simply haven't thought very much about the impact of AI on economic growth." A quick skim suggests the sample selection was for economists actively working on the effects of AI. 

I also think 3.5% growth is under-ratedly big. Absent AI, my guess is that most economists would predict a growth slowdown (demographic drag, ideas getting harder to find, etc.) The counterfactual rate could be something like ~1.75 in 2050. If so -- this implies rapid AI progress would 2x the rate of economic growth relative to no AI. That's a big deal! 

Very cool stuff. Thanks for writing it up! 

I'm not sure that empirical data point about families is as telling as you think it is. Emissions are tied to total economic production. If total economic production doesn't go up, its hard to see how emissions go up much. The story by which an extra baby causes more economic production has to run through something like a demand-constrained economy (eg we're in a recession where more demand really does translate to more production). 

In reality my best guess is that total production is ~unchanged when a new baby is born (or maybe even falls a bit, if a parent pulls back from the labor force), which means the additional purchases of a household with a baby are offset by slightly fewer purchases elsewhere in the economy. (Obviously once that baby is old enough to themselves work, that does increase production and emissions).

Depopulation is Bad

I think this is directionally likely to be true, and I agree with the main claims of the book that Richard nicely summarizes here. That shouldn't be too surprising -- I work with Mike and Dean and helped with a bunch of the research! Thanks for sparking this debate :) 
 

My main uncertainty has to do with a further idea I've been developing (with Loren Fryxell). The thrust of it can be seen with an oversimplified example. Suppose it will take 200 billion human-lives worth of progress to get to the technology that is so dangerous that we go extinct soon after its invention. The effect of adding people today is to accelerate the arrival of this dangerous tech in a way that undoes all of the purported benefits of the earlier people (ie, the exact same number of lives with the exact same average quality are lived). The model we write down is in agreement with essentially all of the important claims in the book, it just layers on an explicit extinction assumption that turns out to generate a negative effect that exactly offsets the positive ones. 

This toy example isn't enough to move me to be completely uncertain about whether depopulation is bad (I put 60% agreement, after all). But we show that additional people in the near-term can be harmful, neutral, or beneficial depending on assumptions made about the shape of the people-progress relationship and the types of existential risks we face. I am not very confident that our world is one where those parameters are such that people today are beneficial--there's too little evidence imo. 

Congrats and good luck!

I have no immediate or useful feedback on this specific question, but just wanted to say that I'll be starting as an economics professor in the Fall at Middlebury. I'd be excited to meet and engage with y'all! If any of your identified bottlenecks are something a faculty member would be able to help with, keep me in mind :) 

Hi Vasco, thanks for reading. And thanks for your dedication to animals :) I've seen a few of your posts on this topic.

If you think you'll be interested in economics PhD programs, I would encourage you to aim to apply for the next cycle (Dec '24/Jan '25). There's a lot of randomness in the process, and your grades will matter more than RA-experience, so I'd say go for it as soon as you can, given how long these programs are. If you don't get in anywhere, you can be applying for RA-ships in the meantime, and take one if that's your best option before trying again the following cycle. You should be able to determine within the next 10 months whether you're interested in the material enough to set off on the PhD, and I wouldn't waste any time applying if you decide you are.

However, I might recommend engaging with economics research directly, rather than Marginal Revolution courses. That will be a better flavor of what you'll do in a PhD. Even if don't understand everything in cutting edge research articles, you'll be able to get a sense for how problems are discussed and debated, which will be a clue as to whether or not its a field you're excited by. (Maybe start here, or here)

Good luck!

I love this post. So much so that I got up to do a set of pull-ups between reading it and writing this comment! 

I'm currently 95% bivalve vegan (occasional eat some cheese and other trace dairy products when I'm out of my house). I'm the closest thing to fully plant based that most of my friends and family know. My jam has been long-distance running rather than getting strong, but some fraction of my motivation has similarly been 'be fitter than almost every meat eater they know.' 

Unfortunately a running physique perpetuates the frail stereotype, so this is a good reminder to do some curls every once in a while for the animals :)

[Oh, and as an N=2 on the substance of this post: I am much healthier and fitter now than I was before going veg, but that's largely confounded by also getting more into running over the last 3 years. The increase in fitness over this period has been significant enough that I'd be surprised if eating meat would have resulted in an even greater increase.] 

At what time horizon? For anything over a year, I'd default to the quantity theory of money: inflation should roughly equal the rate of money supply growth (i.e., a central bank choice) minus real rates of economic growth. Increasing the money supply at 30% per year is easy, so if the Fed wanted to avoid deflation it seems like it could. The short-run during such a dramatic regime change could become whacky.

+1. I found it to be an extremely  thought provoking, informative, and high-quality post. Really well done. [FWIW: I had very weak priors over AGI timelines (I'm too confused to form a coherent inside view) and this seems like a much more reliable outside view than I was defaulting to].