prediction
date: 30/05/2025
market:
Will gross world product in the year 2030 be at least 1 quadrillion USD?
commentary:
By 2030 we'll have ASI. The costs of food, rent, gas, and healthcare will have dropped roughly to zero, so even if GDP is very small the GDP in 2022 dollars will be enormous.
But how sure am I that GDP will be non-zero?
And what about artificial time environments?
GDP: people will be spending their money on... compute. Housing etc. will be free. And people probably won't be spending their money on compute, they'll have their compute allocation and their use of it will be in stasis until such a point as their advisor has reached near-maximum intellect (to ensure wise/optimal use).
Maybe a few people will decide to carry out some purchases, even though that makes no sense, because people may have some measure of freedom, and use it to make poor decisions. This is eminently plausible. Either that or AI just forces everyone to upload / scans everyone's minds and spins them up virtually at some point in the future. In either case GDP is likely zero.
What about non-human GDP?
They could barter but I don't think GDP generally includes that.
All-in-all, a massive GDP doesn't seem that likely.
One might ask "then, what about your stock options expiring in late 2030?", to which I would respond: there will likely be a point sometime in 2027 or 2028 during which each person is apportioned his compute budget; my options will serve me well then.
date: 28/05/2025
market:
In early 2028, will an AI be able to generate a full high-quality movie to a prompt?
commentary:
I saw this video on X. This is strung together by a human over the course of a day or so using Google's recently-released Veo 3 vidgen AI. Two and a half years is a long time in AI, and vidgen seems to be progressing quickly. There are certain known unknowns: when will indefinite shot length be achieved, how quickly will AI be developed to more comprehensively simulate physics, etc. But all of these seem exactly the kinds of problems on which great progress can be quickly made, backed by the full force of Google (as well as many competitors). Also, by 2028 we'll have advanced AGI (at a minimum) which will likely have greatly accelerated the rate of vidgen research. So, YES on this market seems like a good bet. Prob of YES resolution: >0.9