AI 2040, forecasting founders, capital allocation, the end of OpenPhilanthropy's forecasting programme || Forecasting newsletter
What will 2040 be at least as wacky as?
People spend an inordinately large amount of time thinking through the finer points of whether “human-equivalent” intelligence, AGI, ASI, transformative AI, etc. will be reached by 2028, 2030, whenever. But forecasting 2040 or 2100, or overall the long-term trajectory, seems much easier because you can assume diffusion of current technologies and come up with worlds that reality will be “at least as wacky as” in ways which are already actionable and conceptually useful.
Waymo is spreading around the US, but I’m less sure about the rest of the world since roads are worse. LLMs will continue to become better and better, and by 2040 automate a big chunk of current intellectual labor. Widespread robotic helpers seem like they are coming, and there will be huge pressure to use them across Europe since there is capital to buy them and the demographics are terrible; (just look at Spain’s population popsicle). AI tools will diffuse through biochem. The US government will have access to increasingly powerful version of Palantir’s technology. And so on.
I’ve been making a list of things that people can do once big chunks of current labor is automated. Two at the top are stewarding the capital you own, and romance & love. My sense is that the US will further retreat into status and mating displays that will look increasingly insane to the rest of the world.
One of the good futures I can glimpse is creating paradise on Earth and expansion beyond. SpaceX is starting to put datacenters in space. It seems like a stunt so far, but if you extrapolate this 20-200 years it instead seems like the beginning of starting to do cool stuff like colonizing the rest of the solar system, Dyson spheres, expanding through the universe.
I’m pretty confused about how paradise on Earth might look. An ideal would be to just use the ability to transform the world enabled by AI to create really nice environments for humans to live in and flourish, but I’m not sure whether we will get that.
I’m pretty curious about whether US citizens will get an UBI. On the one hand, I don’t see how you avoid this if the democratic system holds, since political power is fairly evenly distributed and in principle the State has a monopoly on violence. But on the other hand if you get transformational technologies, then capital and thus decision-making in a capitalist system becomes more concentrated. This suggests either redistribution or a break in the political order. But the problem with redistribution is that there are a lot of people and you can never have enough, so I’m pretty curious about what kind of balance will be struck here. Even if a balance gets struck within the US, I’m pretty curious about what happens with countries that don’t have sovereign AI capabilities.
Anyways, thinking through what 2040 will be at least as wacky as seems like a productive framework to me.
Forecasting founders I feel excited about
Over the last few years I’ve been compiling lots of factual information about the forecasting ecosystem, but what people have found most interesting and myself most worth writing are the opinions.
I’ve also realized that the outcome of projects has something of a self-fulfilling prophecy, where the more people are excited1 about something, the more likely something might arise as a result of that excitement. Hence there is value in moving people from the lesser fixed point from the greater, and this work is distinct from prediction per se. It’s not like the idea is new, but I’ve internalized it more through personal experience having people believe or not believe in my projects.
In that spirit, here are a few projects I’m excited about and to which I’m happy to direct some attention in the hopes of increasing the probability that they pushing through to that higher point.
Gordon Brander, deepfuture.now: I met Gordon at the Future of Life Fellowship. While I was doing a trip around the US grieving the end of a long-term relationship, he coded an initial scenario planning/strategic foresight tool using AI assistants, which blew my mind. Actionable: check out his app (or join the waitlist). It will ask you a bunch of questions about a scenario you want to explore, and then reward you with an analysis of forces and connections.
Finn Hambly has a really nice forecasting platform up, antistatic.exchange, with some mechanistic improvements that just make it nicer to use. I’m curious about whether it will become its own thing, or whether its alpha will get eaten up by other platforms.
@exgenesis. The guy and his collaborators keep putting out good projects analyzing discourse, centered around a community archive of tweets voluntarily uploaded and the tooling this enables. Actionable: Follow his substack, and check out his underlying institution. Think about how to build cool tooling on top of twitter data.
Nathan Young, has a good feedback loop going on of building stuff and putting it out on the internet; two recent examples are AI-generated community notes on twitter or this dashboard about Iran risk. Actionable: introduce him to people working on community notes at Meta or Tiktok.
@0xperp: The guy behind it is a cracked programmer. Unfortunately his day job pays too much, and so the current platform he has built is just a part time thing. His indices were a good idea. I was hoping he would pull through and build a superior version of my old metaforecast, but this hasn’t happened. If he raises a few million I expect he’d be able to do something great. Actionable: follow the guy on twitter. If you are a VC, inquire about his next round.
A friend started an AGI lab. It might be the beginning of an Anthropic level lab in a few years. It’s been interesting to watch from the sidelines. I’m looking forward to these guys coming out of stealth so that I can cheer for them more publicly.
@niels__ma. I met up with Marc when I was in London, and he struck me as a more suave version of myself (the guy is French). He was previously doing a crypto forecasting aggregation project which has now evolved into an AI aggregation forecasting project. I appreciated his shift as illustrative of how entrepreneurs shift guided by the invisible hand of the market; I knew that there were many people who pivoted from crypto to AI, but I got something out of meeting someone likeable who had gone through that step.
Wasabi pesto was hanging around the forecasting community doing cool projects. Looking forward to the next one.
I’m also excited about my own Sentinel. Perhaps readers might also be interested in some of the intermediary parsing we’ve doing for Reddit (2) and Twitter.
Capital allocation.
Stocks and financial instruments in general are much more liquid than prediction markets, and thus if you have some opinion about the world it makes sense to try to connect it to the financial markets (as opposed to only trying to do things yourself, convincing friends and acquaintances, starting a nonprofit). I’ve been doing some small experiments, making small bets on things that people recommend on twitter, thinking about oil, and listening to my father. I’ve made some small profit, but I’m still very green.
I’ve also been investing in projects in Paraguay, where I’m living. It feels like investing in easy mode, since economic growth is by default 6% a year and so even relatively naïve opportunities should get a decent return (10-20%). But also, at a GDP per capita of $6.4K (vs ~$85K in the US, or $35K in Spain), iteration and learning is 3-20x cheaper. You can get someone decent working for a year for $7.5K, you can buy an apartment for $50K, you can build a small restaurant for $30K, and so on.
And I’ve been trying to acquire small stakes in businesses that my friends and acquaintances start; if I’d started doing that five years ago I could probably have retired by now.
Meanwhile, I’ve been gathering stories of wealth. A Mennonite who is employing 250 people and has 18 gas stations. A real estate broker selling 11 apartments a month for a 5% commission. A residency service making 2k profit from 80 people each month with negligible costs. A friend starting a startup, raising $2M and aiming for $20M more. A guy building up a restaurant chain and catering business with 50 locations. My horizons expand.
The overall theme is that I’ve lived frugally and initially worked in the nonprofit sector, and as I have gotten older I’m going from just deciding how to allocate my own labor to also deciding how to allocate the capital that I’ve been accumulating. I’m in awe at the concept of stock once I start to think about it, because it allows self-reinforcing loops and coalitions around them as opposed to begging people for money2. It’s being an interesting journey.
The connection to the forecasting field is as follows: prediction markets have the nice feature of having a resolution at a given point in time, but they struggle to be connected to real world decisions, and hence they may fall into the temptation of becoming a casino. But stock and financial markets solve the problem of being relevant to decisions by asking the question: “will this guy succeed” in a way that rewards supporters with large, not immediately bounded upside if this happens, as part of a positive sum game. So the connection to the decision is baked-in from the start, solving one of the biggest problems that forecasting and prediction markets historically have had.
The problem was never that policymakers lacked foresight. It was that, once events seemed to validate optimism rather than caution, foresight lost its authority. Butler had sketched the road to escalation in advance; American policy nevertheless followed it.—”Secretary of Defense Rock”, George H. Butler and the Limits of Being Right
Consider signing up for a paid subscription: I’m trying to figure out if I want to continue writting this newsletter and I appreciate the market signal. I’ll do my best to get you your money’s worth of entertainment and information.
Below the paywall are a few words about OpenPhilanthropy’s forecasting grantmaking, I feel a bit more free to speak now since a grant related to this newsletter will not get renewed.