I would love to spend my time just following my curiosity: learning things simply because they fascinate me, taking more time to actually understand them instead of stopping at the point where they're useful enough, being in awe about a question without having to turn it into anything. But I also believe I can currently contribute, so spending my time potentially not contributing would come with (at least a bit of) guilt.

So I landed on this plan: help work toward the point where AI makes my contribution irrelevant, and then live exactly as described, but guilt-free.

The argument

Definitions

• Point X: the state in which AI systems can perform virtually all economically valuable work at least as well as (and at lower cost than) humans. (Most peoples work has a counterfactual value ≈ 0).
• Counterfactual value: the difference an allocation of one's capacity makes to others' good compared to the best available substitute.

Premises

• P1 One ought (prima facie) to allocate one's scarce productive capacities (intellectual energy, skill, ...) where they have the most counterfactual value for others' good.
• P2 The strength of this obligation is proportional to the counterfactual value of one's contribution (therefore: value ≈ 0 → no obligation).
• P3 Point X is achievable.
• P4 One feels rational guilt about not doing A only if one believes (with justification) that one ought to do A.
• P5 A person at point X whose counterfactual value ≈ 0 will know this, accept P1–P2, and therefore not believe they ought to contribute economically.
• P6 Guilt about not doing A diminishes one's enjoyment of doing B (≠ A, curiosity driven activities with potentially no counterfactual value) and enjoying B is good.
• P7 The duties that survive point X (e.g. presence-based, constitutively human goods) don't compete for one's intellectual energy budget.

Theorems

• T1 (from def. of X + P2): At point X, virtually no one is under such obligation.
• T2 (from T1 + P4 + P5): At point X, such people feel no rational guilt about not contributing (given updated social work ethics).
• T3 (from T3 + P6 + P7): Point X makes curiousity following without focus on outcomes possible without guilt.

Conclusion

• C Point X is good in at least this respect: it removes the guilt that comes with not doing a specific thing and therefore frees the intellectual capacity for whatever it desires. (Whether X is good all things considered requires weighing this against its other consequences, obviously)

Corollary

• By P1–P2, obligations are strongest exactly where counterfactual value is highest, which, pre-X, includes work that advances X itself. So: one ought to contribute to X now.

On counterfactual value

Arguably, it is very possible that following your curiosity and understanding things deeply is the highest counterfactual-value use of one's capacities (and productivity propaganda just had us believe otherwise). Einstein would probably have produced more "safe value" working more at his day job, though, as we now know, he produced far greater value following his curiosity. If following curiosity over outcomes is actually higher expected value, we ought to already do that. However: (1) Most wandering produces nothing. For any individual, "I'm like Einstein" is a claim that is probably unjustified. And even if, contrary to (1), the expected-value case can be justified for most people, it will not currently work, because someone has to take care of the everyday things.

This makes X even more desirable though. Post-X, curiosity-following can universalize. Everyone can take the Einstein route, because the alternative of 'regular work' is no longer valuable.