There are some distinctions to be made here. Cryonics obviously provides a better chance to see the future after dying than rotting six feet under.

Yes, but it is less obvious that the chance is large enough to be worth the money.

Regarding retirement investment, just ask your parents or grandparents.

My example was that the type of investments that can be relied upon in coming decades is not accessible that way. For example, many in the US trusted their savings in real estate, which had been trustworthy for generations. And then it wasn’t.

Yet this argument against the necessity of empirical data breaks down at some point. Shaping the Singularity is not on par with having a positive impact on the distant poor. If you claim that predictions and falsifiability are unrelated concepts, that's fine. But to believe some predictions - e.g. a technological Singularity spawned by AGI-seeds capable of superhuman recursive self-improvement - compared to other predictions - e.g. a retirement plan for old age - is not the same.

Yes, there is a difference of degree between the difficulty of figuring out whether mortage-backed securities were as trustworthy as people thought (note that this was less obvious beforehand) and the difficulty of thinking non-nonsensically about the impacts of AI on our future. Nonetheless, they both seem sufficiently difficult that their practice is helped by the explicit study of rationality (so that e.g. heuristics and biases, and probability theory, are explicitly studied by many in finance).