“3. Why impartial altruists should suspend judgment under unawareness” by Anthony DiGiovanni

Audio note: this article contains 135 uses of latex notation, so the narration may be difficult to follow. There's a link to the original text in the episode description. To recap, first, we face an epistemic challenge beyond uncertainty over possible futures. Due to unawareness, we can’t conceive of many relevant futures in the first place, which makes the standard EV framework ill-suited for impartial altruistic decision-making. And second, we can’t trust that our intuitive comparisons of strategies’ overall consequences price in factors we’re unaware of with enough precision. We’ll need to evaluate these consequences with a framework that explicitly accounts for unawareness, that is, unawareness-inclusive expected value (UEV). Here, I’ll more specifically unpack the UEV model, and argue that we can’t compare the UEV of any given strategy with another. If so, we don’t have a reason to choose one strategy over another based purely on [...] ---Outline:(01:33) Unawareness-inclusive expected value (UEV)(05:46) Comparing strategies under the UEV model(07:48) Reasons to suspend judgment on comparisons of strategies' UEV(09:18) Problem 1: Modeling the catch-all, and biased sampling(14:18) Problem 2: Severe underspecification of coarse hypotheses(17:33) Case study revisited: Sign indeterminacy of AI safety interventions due to unawareness(18:20) What we're missing: Outcome robustness and implementation robustness(20:45) The problem in detail(26:59) Appendix B: Formalism for unawareness-inclusive expected value(28:42) Appendix C: Incomparability and positive interventions(33:29) ReferencesThe original text contained 8 footnotes which were omitted from this narration. --- First published: June 2nd, 2025 Source: https://forum.effectivealtruism.org/posts/rec3E8JKa7iZPpXfD/3-why-impartial-altruists-should-suspend-judgment-under --- Narrated by TYPE III AUDIO. ---Images from the article:Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.